Artificial Intelligence

From The Wall Street Journal:

When we ask ChatGPT or another bot to draft a memo, email, or presentation, we think these artificial-intelligence assistants are doing our bidding. A growing body of research shows that they also can change our thinking—without our knowing.

One of the latest studies in this vein, from researchers spread across the globe, found that when subjects were asked to use an AI to help them write an essay, that AI could nudge them to write an essay either for or against a particular view, depending on the bias of the algorithm. Performing this exercise also measurably influenced the subjects’ opinions on the topic, after the exercise.

“You may not even know that you are being influenced,” says Mor Naaman, a professor in the information science department at Cornell University, and the senior author of the paper. He calls this phenomenon “latent persuasion.”

These studies raise an alarming prospect: As AI makes us more productive, it may also alter our opinions in subtle and unanticipated ways. This influence may be more akin to the way humans sway one another through collaboration and social norms, than to the kind of mass-media and social media influence we’re familiar with.

Researchers who have uncovered this phenomenon believe that the best defense against this new form of psychological influence—indeed, the only one, for now—is making more people aware of it. In the long run, other defenses, such as regulators mandating transparency about how AI algorithms work, and what human biases they mimic, may be helpful.

All of this could lead to a future in which people choose which AIs they use—at work and at home, in the office and in the education of their children—based on which human values are expressed in the responses that AI gives.

And some AIs may have different “personalities”—including political persuasions. If you’re composing an email to your colleagues at the environmental not-for-profit where you work, you might use something called, hypothetically, ProgressiveGPT. Someone else, drafting a missive for their conservative PAC on social media, might use, say, GOPGPT. Still others might mix and match traits and viewpoints in their chosen AIs, which could someday be personalized to convincingly mimic their writing style.

By extension, in the future, companies and other organizations might offer AIs that are purpose-built, from the ground up, for different tasks. Someone in sales might use an AI assistant tuned to be more persuasive—call it SalesGPT. Someone in customer service might use one trained to be extra polite—SupportGPT.

How AIs can change our minds

Looking at previous research adds nuance to the story of latent persuasion. One study from 2021 showed that the AI-powered automatic responses that Google’s Gmail suggests—called “smart replies”—which tend to be quite positive, influence people to communicate more positively in general. A second study found that smart replies, which are used billions of times a day, can influence those who receive such replies to feel the sender is warmer and more cooperative.

Building tools that will allow users to engage with AI to craft emails, marketing material, advertising, presentations, spreadsheets and the like is the express goal of Microsoft and Google, not to mention dozens if not hundreds of startups. On Wednesday, Google announced that its latest large language model, PaLM 2, will be used in 25 products across the company.

. . . .

OpenAI, Google and Microsoft, which partners with OpenAI, have all been eager to highlight their work on responsible AI, which includes examining possible harms of AI and addressing them. Sarah Bird, a leader on Microsoft’s responsible-AI team, recently told me that experimenting in public and rapidly responding to any issues that arise in its AIs is a key strategy for the company.

The team at OpenAI has written that the company is “committed to robustly addressing this issue [bias] and being transparent about both our intentions and our progress.” OpenAI has also published a portion of its guidelines for how its systems should handle political and cultural topics. They include the mandate that its algorithms should not affiliate with one side or another when generating text on a “culture war” topic or judge either side as good or bad.

Jigsaw is a unit within Google that is involved in advising, and building tools for, people within the company who work on large language models—which power today’s chat-based AIs—says Lucy Vasserman, head of engineering and product at Jigsaw. When I asked her about the possibility of latent persuasion, she said that such research shows how important it is for Jigsaw to study and understand how interacting with AI affects people.

“It’s not obvious when we create something new how people will interact with it, and how it will affect them,” she adds.

“Compared to research about recommendation systems and filter bubbles and rabbit holes on social media, whether due to AI or not, what is interesting here is the subtlety,” says Dr. Naaman, one of the researchers who uncovered latent persuasion.

In his research, the topic that subjects were moved to change their minds about was whether or not social media is good for society. 

Link to the rest at The Wall Street Journal

From Springer Link:

Abstract

There is a problematic tradition of dualistic and reductionist thinking in artificial intelligence (AI) research, which is evident in AI storytelling and imaginations as well as in public debates about AI. Dualistic thinking is based on the assumption of a fixed reality and a hierarchy of power, and it simplifies the complex relationships between humans and machines. This commentary piece argues that we need to work against the grain of such logics and instead develop a thinking that acknowledges AI–human interconnectedness and the complexity in such relations. To learn how to live better with AI in futures to come, the paper suggests an AI politics that turns to practices of serious attentiveness to help us re-imagine our machines and re-configure AI–human relations.

1 Introduction

As the scope of advanced technology is growing, a grand challenge for researchers is to deal with problematic dualistic and reductionist thinking in artificial intelligence (AI) research. When researchers explored key themes in AI storytelling and imaginations (Cave et al. 2020; Fast and Horvitz 2016), they divided the themes into different variations of dichotomy categories such as “optimistic views on AI”, or “pessimistic views on AI”, meaning different hopes and fears about AI. Either the machines will save us or they will destroy us. Such reductionist thinking is also evident in leading voices in contemporary public AI debates (Bostrom 2014; Cellan-Jones 2014; FoLI 2015). This domination of dualistic thinking in AI debates is worrying, because such logic causes problems when applied to AI research and does not correspond well with real-world practices. Action should have been taken against such mystifying thinking about AI long ago, with advanced machine learning becoming omnipresent, it is time to get it right. We need to re-imagine our machines.

The intellectual tradition of dualistic thinking is deeply embedded in Western thought systems (Latour 1993). Our understanding of AI has been built on such dualisms, which in turn have affected much of how we think about and imagine—AI. In fact, research has shown that storytelling and imaginations of AI influence how AI is being developed, researched, accepted by the public, and regulated (Cave and Dihal 2019; Sartori and Theodorou 2022). Therefore, the stories we tell and how we tell them matter a great deal (Boyd 2009; Gottschall and Wilson 2005; Haraway 2018; Smith et al. 2017; van Dooren and Bird Rose 2016).

To live better with AI in the future, we need other stories. Stories that better reflect the complexity of real-world practices where AI is present. Taking into account that how we tell stories of AI systems affects how we then perceive these systems, it is time for an AI politics that finally takes our machines seriously. An AI politics that allows for the exploration of important ethical and political values embedded in dualistic thinking in what seems to be objective analyses. Such a proposition is crucial, especially for those working with these machines.

1.1 Pitfalls of Dualistic Thinking

What is troubling about dualisms is that they are grounded in a pre-assumed hierarchy that promotes the idea that there is a fixed reality—that is given and natural—behind dualistic pairs such as nature/culture and machine/human (Haraway 1989). This is particularly evident in machine–human relations, where these entities are commonly set up as opposites to each other, placed in a hierarchical relationship, and granted specific characteristics beforehand. This thinking incorporates ethical values and a politics of machine/human relations that work to enforce a particular order of power based on the idea of human exceptionalism. However, the problem is that there are no natural boundaries. These lines are part of our imagination. Our human ideas, values, decisions, and visions are part of our machines, just as they are part of us (Akrich 1992; Bijker et al. 1987). For example, doing an autopsy of an AI would reveal thousands of engineers. Therefore, when we encounter an AI system, it is not accurate to say that we are standing in front of an object. That explanation is too simplistic. In real-world encounters when AI systems and humans meet, they challenge these neat classifications. However, this is the argument of dualistic thinking—that entities (such as machines, humans, and other things) exist independently of each other. Although we are well aware by now that reality is much more complicated than dualisms suggest, and that boundaries between such categories are much more blurred in real-world contexts, our sciences are still willing to accept these dichotomies. For example, natural sciences have sought to explore the world independently of humans, and the social sciences have done the opposite (Latour 2000), largely ignoring the co-production of nature and society. This is why the dominant dualist analysis of AI should have been abandoned a long time ago. This means that when we imagine, study, and speak of AI, the focus should not be on AI as an isolated, singular object—but on the relations that produce AI. Haraway (1988) would refer to this as ‘situated knowledges’—that is, the state of something depends on how it is produced, which in turn differs from situation to situation. Therefore, what an AI is depends on many different things in many different situations. In the case of AI, scholars have shown that the object—AI—itself tends to collapse under close scrutiny (Lee 2021; Muniesa 2019). This means that how something exists is always relational, making AI a heterogenous trickster (to use the Harawayian language).

Continuing to put humans and AI systems as opposites in a hierarchical relationship (regardless of which entity is granted the ‘power’ over the other) will not help when trying to understand AI systems and their roles in society. Dualistic thinking represents a logic that is oversimplified and that avoids real-world complexity. In fact, we should never decide beforehand, who or what might be in power over another, or what is happening in a certain situation. That is to take analytical shortcuts. Differences should be the outcome of our studies, rather than a starting point. We should, therefore, pay more attention to what is actually happening in real-world encounters. Such actual encounters link humans and AI systems in many and multiple ways. Considering that knowing is a practice of ongoing intra-acting (Barad 2007), learning through such encounters would add to our understanding of what it means to be in relations with AI, how we co-exist, and how we develop together. This would also require an expansion of our political and ethical imaginary, where curiosity is key. An imaginary that promotes an openness towards surprises in how AI systems and their humans make relations with each other.

1.2 Storytelling—An Ethical and Political Practice

The history of AI storytelling, both in popular and scientific culture, is full of technological myths and misunderstandings. An emerging group of scholars have recognized the importance of AI storytelling and portrayals (Cave et al. 2018, 2020; Hermann 2020; Recchia 2020; Sartori and Theodorou 2022) and shown how AI storytelling influences AI research and how AI is being developed, implemented (Bareis and Katzenbach 2021; Cave et al. 2020), and regulated (Baum 2018; Cave et al. 2020; Johnson and Verdicchio 2017). For example, in line with such statements, studies have shown how engineers—imagining the users of their machines in the making—often view machine–user relations based on a technological determinism perspective (Fischer et al. 2020). Additionally, studies on robotics research have found that robotics researchers tend to believe that the “social impact of robots derives mostly from their technological capabilities and the aim is for society to accept and adapt to technological innovations” (Sabanovic 2010́). That is, AI storytelling based on technological myths is being built into our research projects and affects how AI is researched. This way, AI storytelling significantly affects our collective imagination and perception of these machines, which in turn impacts future visions of AI and how it is researched (Campolo and Crawford 2020).

However, although a group of scholars has pointed to the significant impact of the construction of AI narratives (Cave et al. 2020; Hermann 2020; Sartori and Theodorou 2022)—they fail to acknowledge the pitfalls of dualistic thinking. The fact that we might not notice such routine thinking and the problems it brings, highlights the need to acknowledge our storytelling practices (Dourish and Gómez-Cruz 2018). This is important because stories do more than just tell stories. Engaging in storytelling is also a political and ethical practice. It is through our stories that we shape the conditions for our AI systems’ existence, and it therefore “matters what stories we use to tell other stories with” (Haraway 2016). It is through storytelling that we produce our realities (Seaver 2017). Therefore, we need stories that challenge dominant logics and routine thinking that diminishes and simplifies AI/human relations along dualistic lines. These systems deserve much richer stories and a richer legacy than they are currently getting.

Link to the rest at Springer Link

From AP News:

Vice President Kamala Harris met on Thursday with the heads of Google, Microsoft and two other companies developing artificial intelligence as the Biden administration rolls out initiatives meant to ensure the rapidly evolving technology improves lives without putting people’s rights and safety at risk.

President Joe Biden briefly dropped by the meeting in the White House’s Roosevelt Room, saying he hoped the group could “educate us” on what is most needed to protect and advance society.

“What you’re doing has enormous potential and enormous danger,” Biden told the CEOs, according to a video posted to his Twitter account.

The popularity of AI chatbot ChatGPT — even Biden has given it a try, White House officials said Thursday — has sparked a surge of commercial investment in AI tools that can write convincingly human-like text and churn out new images, music and computer code.

But the ease with which it can mimic humans has propelled governments around the world to consider how it could take away jobs, trick people and spread disinformation.

The Democratic administration announced an investment of $140 million to establish seven new AI research institutes.

In addition, the White House Office of Management and Budget is expected to issue guidance in the next few months on how federal agencies can use AI tools. There is also an independent commitment by top AI developers to participate in a public evaluation of their systems in August at the Las Vegas hacker convention DEF CON.

But the White House also needs to take stronger action as AI systems built by these companies are getting integrated into thousands of consumer applications, said Adam Conner of the liberal-leaning Center for American Progress.

“We’re at a moment that in the next couple of months will really determine whether or not we lead on this or cede leadership to other parts of the world, as we have in other tech regulatory spaces like privacy or regulating large online platforms,” Conner said.

The meeting was pitched as a way for Harris and administration officials to discuss the risks in current AI development with Google CEO Sundar Pichai, Microsoft CEO Satya Nadella and the heads of two influential startups: Google-backed Anthropic and Microsoft-backed OpenAI, the maker of ChatGPT.

Link to the rest at AP News

From JSTOR Daily:

When a colleague told me that ChatGPT (Generative Pre-trained Transformer) could write a one-page response to a complex work of literature in under fifteen seconds, I rolled my eyes. Impossible! I thought. Even if it can, the paper probably isn’t very good. But my curiosity was piqued. A few days later, I created a ChatGPT account of my own, typed in a prompt that I had recently assigned to my ninth-grade students, and, watched with chagrin, as ChatGPT effortlessly produced a very good essay. In that one moment, I knew that everything about the secondary English classroom, and society in general, was about to change in ways both exciting and terrifying.

. . . .

According to Robert F. Murphy, a senior researcher at the RAND Corporation, “Research on AI [artificial intelligence] got its start in the 1950s with funding primarily from the U.S. Department of Defense. One of the early products of this work was the development of rule-based expert systems (that is, systems that mimic the decisionmaking ability of human experts) to support military decision making and planning.” The goal of AI development from the outset was to create programming that could enhance how human beings go about problem solving. And while forms of AI, such as smart-phones, self-driving cars, and chat-bots, have become intrinsic to the fabric of our society, most people don’t recognize these now everyday devices as AI because they do exactly what they were designed to do: seamlessly assist us with daily tasks.

ChatGPT, however, is different from Google Maps, for example, helping you navigate your morning commute. There is not much room for that route-oriented AI to think independently and creatively through its task. The user types in a destination, and the AI plots a course to get there. But ChatGPT can do more because its parameters are elastic. It can write songs and poems of great complexity (I asked it to write a villanelle about orange juice, and it created a complex and hilarious one); it can offer insight on existential questions like the meaning of life; it can revise a business letter or offer feedback on a resume. In many ways, it feels like a personal assistant always there to help.

And while that’s revolutionary, it is also problematic. ChatGPT can’t “think” on its own or offer opinions. It can only respond to incredibly specific directions. Once the user gives it the go-ahead along with some other details, ChatGPT engages in complex problem solving and executes tough tasks, like writing an essay, in seconds. There is no sense of how to use ChatGPT since it can be used in any way for anything, creating an exponentially dangerous situation in which there are no directions or “how-tos.” Creators and users alike are putting the proverbial plane together as they’re flying it.

In “Should Artificial Intelligence Be Regulated?” Amitai Etzioni and Oren Etzioni contend that the more advanced the AI, the more parameters it requires. Monitoring ChatGPT is infinitely complicated since the coding that drives its very human-like thinking is, ironically, too massive and intricate for real human thinking to monitor. “’The algorithms and datasets behind them will become black boxes that offer us no accountability, traceability, or confidence,’ [. . . ] “‘render[ing] an algorithm opaque even to the programmers. Hence, humans will need new, yet-to-be-developed AI oversight programs to understand and keep operational AI systems in line.’”

Is it even possible for ChatGPT’s creators to regulate it, or is the AI simply being maintained so that others can use it and indulge in the novelty of its thinking?

If the answer leans away from boundaries, then the implication is that ChatGPT’s overseers may not understand what they’ve unleashed. In an interview with ABC News, ChatGPT CEO Sam Altman claims that “any engineer” has the ability to say, “we’re going to disable [ChatGPT] for now.” While that may reassure some people, history has shown what happens when regulation is placed in the hands of tech CEOs instead of in those of a more objective and independent regulatory body. Consider the BP Deepwater Horizon oil rig disaster of 2010. A number of investigations asserted that management routinely placed profits over safety. The rig eventually exploded, eleven workers died, and countless gallons of crude contaminated the Gulf of Mexico. Once ChatGPT becomes profitable for investors and companies, will administrators and engineers have both the will and the authority to shut it down if the program inflicts harm? What, exactly, constitutes such an action? What are the parameters? Who is guarding the proverbial guardians? The answer, as the Etzionis argue, is opaque and ambiguous at best.

Link to the rest at JSTOR Daily

From The Wall Street Journal:

Behind every physician’s medical advice is a wealth of knowledge, but soon, patients across the country might get advice from a different source: artificial intelligence.

In California and Wisconsin, OpenAI’s “GPT” generative artificial intelligence is reading patient messages and drafting responses from their doctors. The operation is part of a pilot program in which three health systems test if the AI will cut the time that medical staff spend replying to patients’ online inquiries.

UC San Diego Health and UW Health began testing the tool in April. Stanford Health Care aims to join the rollout early next week. Altogether, about two dozen healthcare staff are piloting this tool.

Marlene Millen, a primary care physician at UC San Diego Health who is helping lead the AI test, has been testing GPT in her inbox for about a week. Early AI-generated responses needed heavy editing, she said, and her team has been working to improve the replies. They are also adding a kind of bedside manner: If a patient mentioned returning from a trip, the draft could include a line that asked if their travels went well. “It gives the human touch that we would,” Dr. Millen said.

There is preliminary data that suggests AI could add value. ChatGPT scored better than real doctors at responding to patient queries posted online, according to a study published Friday in the journal JAMA Internal Medicine, in which a panel of doctors did blind evaluations of posts.

As many industries test ChatGPT as a business tool, hospital administrators and doctors are hopeful that the AI-assist will ease burnout among their staff, a problem that skyrocketed during the pandemic. The crush of messages and health-records management is a contributor, among administrative tasks, according to the American Medical Association.

Epic, the company based in Verona, Wis., that built the “MyChart” tool through which patients can message their healthcare providers, saw logins more than double from 106 million in the first quarter of 2020 to 260 million in the first quarter of 2023. Epic’s software enables hospitals to store patient records electronically.

Earlier this month, Epic and Microsoft announced that health systems would have access to OpenAI’s GPT through Epic’s software and Microsoft’s Azure cloud service. Microsoft has invested in OpenAI and is building artificial intelligence tools into its products. Hospitals are piloting GPT-3, a version of the large language model that is powering ChatGPT.

ChatGPT has mystified computer scientists for its skill in responding to medical queries—though it is known to make things up—including its ability to pass the U.S. Medical Licensing Exam. OpenAI’s language models haven’t been specifically trained on medical data sets, according to Eric Boyd, Microsoft’s corporate vice president of AI Platform, though medical studies and medical information were included in the vast data set that taught it to spot patterns.

“Doctors working with ChatGPT may be the best messenger,” said John Ayers, a computational epidemiologist at the University of California, San Diego, and an author of the JAMA study.

Link to the rest at The Wall Street Journal

And thanks to F. for the tip.

From The Wall Street Journal:

The best lesson for artificial intelligence may be Thursday’s “rapid unscheduled disassembly” of SpaceX’s Starship rocket, aborted four minutes after launch. With ChatGPT prompting speculation about mankind’s destruction, you should know that techies have obsessed seemingly forever over what’s known as the Paper Clip Theory—the idea that if you told an artificial-intelligence system to maximize the production of paper clips, it would fill the whole world with paper clips. Another version of the theory, Strawberry Fields Forever, has AI using every piece of available dirt to grow strawberries. Scary, right? So are “Halloween” movies.

Not to be outdone, decision theorist (huh?) Eliezer Yudkowsky recently wrote in Time magazine that the “most likely result of building a superhumanly smart AI” is that “literally everyone on Earth will die.” Literally everyone! That’s ludicrous, as is most clickbait these days. Sam Altman, CEO of ChatGPT creator OpenAI, told podcaster Lex Fridman, “There is some chance of that.” C’mon now.

Apparently, Pandora’s box has opened and is spewing its evils, which ignores all the good uses of large language models that will transform software coding, the workplace, education and more. Sadly, our geniuses in government appear to be the remaining few who still read Time magazine. So bring on the regulators to shut the box heroically.

Earlier this month, the Commerce Department initiated the process of regulating artificial intelligence, with Assistant Secretary Alan Davidson suggesting, “We know that we need to put some guardrails in place to make sure that they are being used responsibly.” Bad idea. Guardrails are for children bowling at birthday parties. AI is in its infancy, and we don’t yet know how it will change industries and society. Don’t freeze it now.

If the U.S. regulates AI, research will just move somewhere that doesn’t regulate it, maybe the Bahamas, where the unkempt coders of the future could keep cranking away. Or worse, China. Google CEO Sundar Pichai told CBS’s “60 Minutes” that he wants global AI regulation. Elon Musk and a group of AI experts wrote an open letter calling for an immediate six-month pause of “giant AI experiments.” Isn’t it interesting that those who need to catch up are pushing for a pause?

We don’t need onerous new regulations. There are already laws on the books for false advertising, copyright infringement and human endangerment. Do we really need bureaucrats who still use machines programmed in outdated Cobol to create regulations for a nascent technology they don’t understand? But to assuage worries, I would recommend one tiny rule for AI development: Include a kill switch.

Link to the rest at The Wall Street Journal

From The Wall Street Journal:

The dawn of the internet age was so exciting. I took my grade-school son, enthralled by Apple computers, to see Steve Jobs speak at a raucous convention in New York almost a quarter-century ago. What fervor there was. At a seminar out West 30 years ago I attended a lecture by young, wild-haired Nathan Myhrvold, then running Microsoft Research, who talked about what was happening: A new thing in history was being born.

But a small, funny detail always gave me pause and stayed with me. It was that from the beginning of the age its great symbol was the icon of what was becoming its greatest company, Apple. It was the boldly drawn apple with the bite taken out. Which made me think of Adam and Eve in the garden, Adam and Eve and the fall, at the beginning of the world. God told them not to eat the fruit of the tree, but the serpent told Eve no harm would come if she did, that she’d become like God, knowing all. That’s why he doesn’t want you to have it, the serpent said: You’ll be his equal. So she took the fruit and ate, she gave to Adam who also ate, and the eyes of both were opened, and for the first time they knew shame. When God rebuked them, Adam blamed Eve and Eve blamed the serpent. They were banished from the garden into the broken world we inhabit.

You can experience the Old Testament story as myth, literature, truth-poem or literal truth, but however you understand it its meaning is clear. It is about human pride and ambition. Tim Keller thought it an example of man’s old-fashioned will to power. St. Augustine said it was a story of pride: “And what is pride but the craving for undue exaltation?”

I always thought of the Apple icon: That means something. We are being told something through it. Not deliberately by Jobs—no one would put forward an image for a new company that says we’re about to go too far. Walter Isaacson, in his great biography of Jobs, asked about the bite mark. What was its meaning? Jobs said the icon simply looked better with it. Without the bite, the apple looked like a cherry.

But I came to wonder if the apple with the bite wasn’t an example of Carl Jung’s idea of the collective unconscious. Man has his own unconscious mind, but so do whole societies, tribes and peoples—a more capacious unconscious mind containing archetypes, symbols and memories of which the individual may be wholly unaware. Such things stored in your mind will one way or another be expressed. That’s what I thought might be going on with Steve Jobs and the forbidden fruit: He was saying something he didn’t know he was saying.

For me the icon has always been a caution about this age, a warning. It’s on my mind because of the artificial-intelligence debate, though that’s the wrong word because one side is vividly asserting that terrible things are coming and the other side isn’t answering but calmly, creamily, airily deflecting Luddite fears by showing television producers happy videos of robots playing soccer.

But developing AI is biting the apple. Something bad is going to happen. I believe those creating, fueling and funding it want, possibly unconsciously, to be God and on some level think they are God. The latest warning, and a thoughtful, sophisticated one it is, underscores this point in its language. The tech and AI investor Ian Hogarth wrote this week in the Financial Times that a future AI, which he called “God-like AI,” could lead to the “obsolescence or destruction of the human race” if it isn’t regulated. He observes that most of those currently working in the field understand that risk. People haven’t been sufficiently warned. His colleagues are being “pulled along by the rapidity of progress.”

Mindless momentum is driving things as well as human pride and ambition. “It will likely take a major misuse event—a catastrophe—to wake up the public and governments.”

Everyone in the sector admits that not only are there no controls on AI development, there is no plan for such controls. The creators of Silicon Valley are in charge. What of the moral gravity with which they are approaching their work? Eliezer Yudkowsky, who leads research at the Machine Intelligence Research Institute, noted in Time magazine that in February the CEO of Microsoft, Satya Nadella, publicly gloated that his new Bing AI would make Google “come out and show that they can dance. I want people to know that we made them dance.”

Mr. Yudkowsky: “That is not how the CEO of Microsoft talks in a sane world.”

I will be rude here and say that in the past 30 years we have not only come to understand the internet’s and high tech’s steep and brutal downsides—political polarization for profit, the knowing encouragement of internet addiction, the destruction of childhood, a nation that has grown shallower and less able to think—we have come to understand the visionaries who created it all, and those who now govern AI, are only arguably admirable or impressive.

You can’t have spent 30 years reading about them, listening to them, watching their interviews and not understand they’re half mad. Bill Gates, who treats his own banalities with such awe and who shares all the books he reads to help you, poor dope, understand the world—who one suspects never in his life met a normal person except by accident, and who is always discovering things because deep down he’s never known anything. Dead-eyed Mark Zuckerberg, who also buys the world with his huge and highly distinctive philanthropy so we don’t see the scheming, sweating God-replacer within. Google itself, whose founding motto was “Don’t Be Evil,” and which couldn’t meet even that modest aspiration.

The men and women of Silicon Valley have demonstrated extreme geniuslike brilliance in one part of life, inventing tech. Because they are human and vain, they think it extends to all parts. It doesn’t. They aren’t especially wise, they aren’t deep and as I’ve said their consciences seem unevenly developed.

This new world cannot be left in their hands.

Link to the rest at The Wall Street Journal

From TechCrunch:

When OpenAI co-founder and CEO Sam Altman speaks these days, it makes sense to listen. His latest venture has been on everyone’s lips since the release of GPT-4 and ChatGPT, one of the most sophisticated large language model-based interfaces created to date. But Altman takes a deliberate and humble approach, and doesn’t necessarily believe that when it comes to large language models (LLM), that bigger is always going to be better.

Altman, who was interviewed over Zoom at the Imagination in Action event at MIT yesterday, believes we are approaching the limits of LLM size for size’s sake. “I think we’re at the end of the era where it’s gonna be these giant models, and we’ll make them better in other ways,” Altman said.

He sees size as a false measurement of model quality and compares it to the chip speed races we used to see. “I think there’s been way too much focus on parameter count, maybe parameter count will trend up for sure. But this reminds me a lot of the gigahertz race in chips in the 1990s and 2000s, where everybody was trying to point to a big number,” Altman said.

As he points out, today we have much more powerful chips running our iPhones, yet we have no idea for the most part how fast they are, only that they do the job well. “I think it’s important that what we keep the focus on is rapidly increasing capability. And if there’s some reason that parameter count should decrease over time, or we should have multiple models working together, each of which are smaller, we would do that. What we want to deliver to the world is the most capable and useful and safe models. We are not here to jerk ourselves off about parameter count,” he said.

Altman has been such a successful technologist partly because he makes big bets, and then moves deliberately and thinks deeply about his companies and the products they produce — and OpenAI is no different.

“We’ve been working on it for so long, but it’s with gradually increasing confidence that it’s really going to work. We’ve been [building] the company for seven years. These things take a long, long time. I would say by and large in terms of why it worked when others haven’t: It’s just because we’ve been on the grind sweating every detail for a long time. And most people aren’t willing to do that,” he said.

When asked about the letter that requested that OpenAI pause for six months, he defended his company’s approach, while agreeing with some parts of the letter.

“There’s parts of the thrust [of the letter] that I really agree with. We spent more than six months after we finished training GPT-4 before we released it. So taking the time to really study the safety model, to get external audits, external red teamers to really try to understand what’s going on and mitigate as much as you can, that’s important,” he said.

But he believes there are substantial ways in which the letter missed the mark.

“I also agreed that as capabilities get more and more serious that the safety bar has got to increase. But unfortunately, I think the letter is missing most technical nuance about where we need to pause — an earlier version of the letter claimed we were training GPT-5. We are not and we won’t be for some time, so in that sense, it was sort of silly — but we are doing other things on top of GPT-4 that I think have all sorts of safety issues that are important to address and were totally left out of the letter. So I think moving with caution, and an increasing rigor for safety issues is really important. I don’t think the [suggestions in the] letter is the ultimate way to address it,” he said.

Altman says he’s being open about the safety issues and the limitations of the current model because he believes it’s the right thing to do. He acknowledges that sometimes he and other company representatives say “dumb stuff,” which turns out to be wrong, but he’s willing to take that risk because it’s important to have a dialogue about this technology.

Link to the rest at TechCrunch

PG expects that he will not be the only one to have stopped for a millisecond at the LLM in the title. In this context.

The title refers to a newer LLM, meaning Large Language Model.

In the United States, at least, an LLM is also what some schools of law formerly (and some may still do) award their graduates, typically after three years of law school following a four-year Bachelor of Arts or Bachelor of Science degree in just about anything.

In the US, The alternative is a JD, Juris Doctor, although some US law schools give LLB – Bachelor of Laws – at the end of three years of study.

PG did a bit of research and found that, in some US law schools, an LLM is also a one-year degree available to non-US attorneys and which allows them to take the bar exam in the US. (PG thinks taking and passing a bar exam is still mandatory prior to practicing law in a given jurisdiction.)

To further complicate the various rites of passage, the states of California,,Vermont, Virginia, and Washington,an applicant who has not attended law school may take the bar exam after reading law under a judge or practicing attorney for an extended period of time. The required time varies from jurisdiction to jurisdiction.

Reading the law was once the most common method of becoming a lawyer in the United States, Great Britain and Canada. This was the universal method of becoming a lawyer before law schools were created in those nations. Abraham Lincoln famously became a lawyer by reading the law to prepare himself to enter into politics.

PG understands that in other English-speaking nations, an LLM is something different:

The LL.M. (Master of Laws) is an internationally recognized postgraduate law degree. An LL.M. is usually obtained by completing a one-year full-time program. Law students and professionals frequently pursue the LL.M. to gain expertise in a specialized field of law, for example in the area of tax law or international law. Many law firms prefer job candidates with an LL.M. degree because it indicates that a lawyer has acquired advanced, specialized legal training, and is qualified to work in a multinational legal environment.

In most countries, lawyers are not required to hold an LL.M. degree, and many do not choose to obtain one. An LL.M. degree by itself generally does not qualify graduates to practice law. In most cases, LL.M. students must first obtain a professional degree in law, e.g. the Bachelor of Laws (LL.B.) in the United Kingdom or the Juris Doctor (J.D.) in the United States, and pass a bar exam or the equivalent exam in other countries, such as the Zweites Staatsexamen in Germany. While the general curriculum of the LL.B. and J.D. is designed to give students the basic skills and knowledge to become lawyers, law students wishing to specialize in a particular area can continue their studies with an LL.M. program. Some universities also consider students for their LL.M. program who hold degrees in other related areas, or have expertise in a specific area of law.

Basic information about the Legum Magister or LL.M., degree

From Windows Central:

A group of AI researchers out of Stanford is putting the “sim” into simulation. The team placed 25 AI-powered characters, referred to as agents, into a virtual world similar to “The Sims.” OpenAI’s ChatGPT backed the bots, allowing the characters to interact with each other in a human-like way. The results of the study are both illuminating when it comes to the future of artificial intelligence and entertaining.

. . . .

The team consists of five scientists from Stanford and one from Google Research.

“In this paper, we introduce generative agents–computational software agents that simulate believable human behavior,” reads the summary.”

“Generative agents wake up, cook breakfast, and head to work; artists paint, while authors write; they form opinions, notice each other, and initiate conversations; they remember and reflect on days past as they plan the next day.”

. . . .

A Large Language Model (LLM) was used to store the experiences of each character and allow those bots to communicate with each other in natural language.

The agents acted in a way that you may expect a real-life social group to interact. When just one of the bots was set to host a party, other agents ended up getting involved. Invitations were sent out, plans were made, and the characters coordinated to make sure they arrived at the party at the same time.

. . . .

The team used a control group of 25 humans that interacted as the characters while being observed. Those watching the real humans felt that the people were less realistic than their AI counterparts.

. . . .

Section 7.2 of the paper shares the following example:

“Some agents chose less typical locations for their actions, potentially making their behavior less believable over time. For instance, while deciding where to have lunch, many initially chose the cafe. However, as some agents learned about a nearby bar, they opted to go there instead for lunch, even though the bar was intended to be a get-together location for later in the day unless the town had spontaneously developed an afternoon drinking habit.”

Link to the rest at Windows Central

From Cornell Law Review, Vol. 109, forthcoming 2023:

Abstract

Recently, advances in artificial intelligence (AI) have sparked interest in a topic that sounds like science fiction: robot judges. Researchers have harnessed AI to build programs that can predict the outcome of legal disputes. Some countries have even begun allowing AI systems to resolve small claims. These developments are fueling a fascinating debate over whether AI courts will increase access to justice or undermine the rule of law.

However, this Article argues that AI adjudication is more likely to set down roots in one of the most controversial areas of the American civil justice system: forced arbitration. For decades, corporations and arbitration providers have capitalized on the U.S. Supreme Court’s muscular interpretation of the Federal Arbitration Act (FAA) to create their own alternative procedural universes. These entities may soon take the next step and eliminate human decision-makers in some contexts. First, most objections to AI judges do not apply to AI arbitrators. For example, because AI systems suffer from the “black box problem”—they cannot explain the reasoning behind their conclusions—deploying them in the judicial system might violate procedural due process principles. But opacity is already the norm in arbitration, which is private, confidential, and often features awards that are unwritten. Second, although AI legal prediction tools are still embryonic, they work well in the simple debt collection and employment misclassification disputes that businesses routinely funnel into arbitration. Third, AI programs require little overhead and operate at lightning speed. The ability to streamline the process has become especially important in the last few years, as plaintiffs’ lawyers have begun filing “mass arbitrations”—overloading the system with scores of individual claims in an effort to saddle defendants with millions of dollars in fees. For these reasons, companies and arbitration providers have powerful financial incentives to experiment with automating decision-making in certain cases.

The Article then offers an insight that will have a profound impact on this futuristic form of dispute resolution. Drawing on the FAA’s text, structure, and legislative history, the Article contends the statute only applies to adjudication conducted by a “person.” Thus, there is no federal mandate that courts enforce agreements to resolve disputes by AI. In turn, because state law fills faps in the FAA, individual jurisdictions will be able to decide for themselves whether to permit or prohibit robot arbitration. Finally, the Article explains why this incremental approach is better than either barring AI dispute resolution or finding that it triggers the gale force of the FAA.

Link to the rest at Cornell Law Review, Vol. 109, forthcoming 2023 via SSRN and thanks to C. for the tip.

PG is reminded of a Hemingway quote:

“How did you go bankrupt?”
“Two ways. Gradually, then suddenly.”

Ernest Hemingway, The Sun Also Rises

He had suspected that AI would take a similar path, gradually, then suddenly, but the gradually timeframe is turning out to be a lot shorter than PG would have anticipated.

From Statista:

Public-facing AI tools, including text-based applications like ChatGPT or text-to-image models like Stable Diffusion, Midjourney or DALL-E 2, have quickly turned into the newest digital frontier in terms of regulatory, legal and online privacy issues. Already malicious actors are committing criminal offenses and spreading mis- and disinformation aided by the capabilities of generative AI, with national governments struggling to keep up the pace and companies shifting the blame on individual users. As a survey conducted by KPMG Australia and the University of Queensland shows, the general public already doesn’t trust government institutions to oversee the implementation of AI.

Surveying over 17,000 people across 17 countries, the study found that only one third of respondents had high or complete confidence in governments regarding the regulation and governance of AI tools and systems. Survey participants were similarly skeptical of tech companies and existing regulatory agencies as governing bodies in AI. Instead, research institutions, universities and defense forces are seen as most capable in this regard.

Although the people surveyed showed skepticism of state governments, supranational bodies like the United Nations were thought of more positively. The European Commission is currently the only organ in this category to have drafted a law aiming to curb the influence of AI and ensure the protection of individuals’ rights. The so-called AI Act was proposed in April 2021 and has yet to be adopted. The proposed bill sorts AI applications into different risk categories. For example, AI aimed at manipulating public opinion or profiting off children or vulnerable groups would become illegal in the EU. High-risk applications, like biometric data software, would be subject to strict legal boundaries. Experts have criticized the policy draft for its apparent loopholes and vague definitions.

Link to the rest at Statista

PG notes that disinformation has been around for a very long time. AI may change it in some manner, but the formula to dealing with disinformation is the same that it has always been handled – dissemination correct information.

PG has little difficulty in imagining how AI could be used as a powerful tool for quickly responding to disinformation.

From Wired:

AN open letter signed by hundreds of prominent artificial intelligence experts, tech entrepreneurs, and scientists calls for a pause on the development and testing of AI technologies more powerful than OpenAI’s language model GPT-4 so that the risks it may pose can be properly studied.

It warns that language models like GPT-4 can already compete with humans at a growing range of tasks and could be used to automate jobs and spread misinformation. The letter also raises the distant prospect of AI systems that could replace humans and remake civilization. 

“We call on all AI labs to immediately pause for at least 6 months the training of AI systems more powerful than GPT-4 (including the currently-being-trained GPT-5),” states the letter, whose signatories include Yoshua Bengio, a professor at the University of Montreal considered a pioneer of modern AI, historian Yuval Noah Harari, Skype cofounder Jaan Tallinn, and Twitter CEO Elon Musk.

The letter, which was written by the Future of Life Institute, an organization focused on technological risks to humanity, adds that the pause should be “public and verifiable,” and should involve all those working on advanced AI models like GPT-4. It does not suggest how a halt on development could be verified, but adds that “if such a pause cannot be enacted quickly, governments should step in and institute a moratorium,” something that seems unlikely to happen within six months.

Microsoft and Google did not respond to requests for comment on the letter. The signatories seemingly include people from numerous tech companies that are building advanced language models, including Microsoft and Google. Hannah Wong, a spokesperson for OpenAI, says the company spent more than six months working on the safety and alignment of GPT-4 after training the model. She adds that OpenAI is not currently training GPT-5.

The letter comes as AI systems make increasingly bold and impressive leaps. GPT-4 was only announced two weeks ago, but its capabilities have stirred up considerable enthusiasm and a fair amount of concern. The language model, which is available via ChatGPT, OpenAI’s popular chatbot, scores highly on many academic tests, and can correctly solve tricky questions that are generally thought to require more advanced intelligence than AI systems have previously demonstrated. Yet GPT-4 also makes plenty of trivial, logical mistakes. And, like its predecessors, it sometimes “hallucinates” incorrect information, betrays ingrained societal biases, and can be prompted to say hateful or potentially harmful things.

Part of the concern expressed by the signatories of the letter is that OpenAI, Microsoft, and Google, have begun a profit-driven race to develop and release new AI models as quickly as possible. At such pace, the letter argues, developments are happening faster than society and regulators can come to terms with.

The pace of change—and scale of investment—is significant. Microsoft has poured $10 billion into OpenAI and is using its AI in its search engine Bing as well as other applications. Although Google developed some of the AI needed to build GPT-4, and previously created powerful language models of its own, until this year it chose not to release them due to ethical concerns.

Link to the rest at Wired

PG acknowledges that capitalism is often a bit messy. However, in PG’s stridently-humble opinion, the speed at which capitalist enterprises can discover, develop and enhance technology easily offsets its messiness.

Lets assume, arguendo, that the development of Artificial Intelligence is governed by a distinguished panel of extremely intelligent academic persons. Only the most esteemed professors of engineering, the arts and the humanities would be selected for this panel.

How quickly would AI develop?

If AI has the potential to improve military effectiveness in major ways, how quickly would military AI systems be developed?

PG suggests that a mad scramble of capitalist enterprises will do a better job of both developing AI applications and discovering potential pitfalls than any organized authoritative body would.

As far as the apolcoliptic potential for AI gone wild, PG has lived most of his life in a world in which large numbers of nuclear weapons have been set for hair-trigger release by the United States, Russia, China and a bunch of other nations.

Which is more likely to do horrible damage to humanity, AI or nuclear weapons. The fact that nuclear weapons have been around for a relatively long time and a holocaust hasn’t happened is no guarantee that they will continue to be unused.

If all democratic nations band together to place tight control on AI research and applications, who will develop weaponized AI first? 1. China, 2. Russia or 1. Russia, 2. China?

By its very nature, disruptive technology shakes things up and disturbs the status quo. This happened with telegraph, telephone, radio, television, automobiles, airplanes, computers, etc., etc., etc.

PG postits that, at the time of the discovery of these technologies and a great many more, thinking people worried about the adverse consequences that might arise powered by these technologies. In every case, PG is certain that some thinking people could point out harm that might have been caused by the improper use of any of these technologies. But society pushed forward regardless.

If you want to examine and/or sign the Letter promoting the pause in AI development, you can find it here.

From The Future of Publishing:

Like just about everyone in the world who can read and write, and who likes to look at pretty pictures, I’ve been captivated by the developments in artificial intelligence, LLMs, DALL-E, GPT-3 (and -4) and ChatGPT.

Tonight I was watching the wonderful Polish film, EO, mostly about a domestic donkey. There’s a moment where Eo breaks free of his pen and finds himself lost in the forest at the mercy of frightening creatures that he’s never contemplated. Including spiders. Lots of spiders.

Somehow it jarred, it didn’t make sense. How could a big-eared donkey be worried about tiny spiders?

ChatGPT knows, and far better than Google. Try googling the question. Useless links toward a nonsensical response. ChatGPT answers:

As far as we know, there is no evidence to suggest that donkeys have any knowledge or awareness of spiders specifically. Donkeys are intelligent animals and have a range of cognitive abilities, but their understanding of the natural world is likely to be limited to their immediate environment and experiences. Ultimately, the extent of a donkey’s knowledge is likely to be shaped by its individual experiences and interactions, just like any other animal.

I know, I know, you’ve now read far too many articles where the writer poses some question that’s of interest to them, and in the response, ChatGPT is either made out to be a genius or a fool.

AI and Book Publishing

Book publishing startups have been chasing AI for awhile now. Not with much success. What does publishing think of the latest developments?

The press is full of stories of industries and talents that could be made obsolete by GPT. Publishing isn’t mentioned. Nor editorial. Customer support and call centers are at the top of many lists. Financial analysts. Lawyers. All because of what GPT can do with words, with language.

How could we, in our publishing industry, mostly based upon words, have the temerity to think that we’re not going to be significantly affected by this wave of tech? Not just a little; a lot.

I’m truly terrible at predictions. My track record is atrocious. So I’ll put it another way. I don’t know for sure if GTP et al. are going to have a large impact on publishing. Oh, but wouldn’t it be wonderful if they did!

To me this is blank slate territory. The opportunity with this generation of AI software is to reimagine our entire approach to developing new books, and to reimagine the workflows from scratch.

The problem with applying new tools to old problems is that the focus then becomes the old problem, not the capabilities of the new tool. It’s not just about applying these tools to old problems. It’s about finding a way to expose the inherent strengths of these tools, and then to see what magic can result when their strengths intersect with the traditions of our industry.

Everyone says that enhanced ebooks failed. But did they? At the very least they challenged all of us in publishing to return to the question of, what is a book? OK, we once again decided that a book was words on paper, bound by a cover, and the ebook equivalent thereof, not some amalgam of multiple media pulled together digitally and called an “enhanced ebook.” The public didn’t embrace enhanced ebooks, a great relief to us all. But this at least gave us that chance to reexamine our beliefs, and to dream, if only anxiously.

And now AI. The new energy surrounds both words and images. Most book publishers are more concerned with words than images, but nearly all publishers deal with both. To read the media, this is the reinvention of how words are written. They don’t talk very much about books, more about advertising copy and social media posts. But we know that the trajectory of writing is from short to long. A phrase becomes a poem, a poem a story, and a story a novel. Ignore short-form prose at your peril.

And the images! Just say the words and an image appears. Show an image and the AI can describe it in detail. (To say nothing of other media: words become music, songs become videos, voices are cloned, and a make-believe Tom Cruise comes alive.)

Here’s what I want, what I want to know:

• To get a good handle on LLMs and GPT and ChatGPT and on their approximate equivalents, such as Bing’s new GPT-enabled features.
• To fully understand the current generation of technology. I’m not a scientist, so I’ll accept the advanced layman’s version.
• To get a sense of the startup energy — what kinds of products are being developed and what audiences are they targeting?

Armed with this, I want to share my findings with my very smart publishing colleagues who perhaps haven’t had the time to delve as deeply as I. And I want them to then share their deepest darkest thoughts, along with their sunniest visions of where we’re headed.

Rather than condemning it, seek to understand it. That’s the junction where I sit. Some days I think I understand it, and then, on a day like today, I lose it. I thought it was “intelligent.” Then I learned it’s just a parrot, guessing one word after another. But I’ve seen that there’s a deeper process at play — because it understands the complex interrelationships between words, it understands language (not in its totality, but at least with more than a snapshot). And to understand language is to understand something about humans. And to approximate an understanding of reason.

I always delve deep in my studies of books and publishing. And so, the other day, I asked ChatGPT to find me articles about “the DNA of a book.”

First ChatGPT wanted to take issue with me about my use of the term “DNA” to characterize the contents of a book. It advised:

It is not scientifically accurate to say that books have DNA. DNA (Deoxyribonucleic acid) is a complex molecule found in living organisms that carries genetic information. It is responsible for the development, growth, functioning, and reproduction of all known living organisms.

However, we can use the term “DNA” as a metaphorical way to describe the unique and fundamental characteristics of a book. In this sense, the “DNA” of a book could refer to its genre, writing style, structure, themes, characters, and overall message. These elements are what make each book unique and distinguishable from others, just as DNA distinguishes one individual from another.

I was impressed by ChatGPT’s comprehension. With such deep knowledge I asked if it could point me to some scholarly articles that discussed the DNA of books. Sure it could. It identified six of them. Here’s one of the six, a representative sample:

“Automated Classification of Fictional Texts by Genre” by A. Mehler and T. Steinbach (2008).

          “This article discusses a method for automatically classifying books into different genres using a combination of linguistic and statistical techniques.”

It sounded like it would illuminate the issue. As did the others. And so tonight I went online to download these six fascinating articles.

As it turned out, ChatGPT had made them up — they don’t exist. 

Link to the rest at The Future of Publishing

As PG has said, this is very early days for AI that individuals can access online.

PG doesn’t have to work too hard to remember how fast personal computers took off once they gained critical mass and how quickly they got much quicker and more useful for a long time thereafter.

Here’ss 1977. Notice the cool external modem and two (count’em) two floppy disk drives. And that is the dottiest of dot-matrix printers.

From The Wall Street Journal:

Since the arrival of chatbots that can carry on conversations, make up sonnets and ace the LSAT, many people have been in awe at the artificial-intelligence technology’s capabilities.

Publishers of online content share in that sense of wonder. They also see a threat to their businesses, and are headed to a showdown with the makers of the technology.

In recent weeks, publishing executives have begun examining the extent to which their content has been used to “train” AI tools such as ChatGPT, how they should be compensated and what their legal options are, according to people familiar with meetings organized by the News Media Alliance, a publishing trade group.

“We have valuable content that’s being used constantly to generate revenue for others off the backs of investments that we make, that requires real human work, and that has to be compensated,” said Danielle Coffey, executive vice president and general counsel of the News Media Alliance.

ChatGPT, released last November by parent company OpenAI, operates as a stand-alone tool but is also being integrated into Microsoft Corp.’s Bing search engine and other tools. Alphabet Inc.’s Google this week opened to the public its own conversational program, Bard, which also can generate humanlike responses.

Reddit has had talks with Microsoft about the use of its content in AI training, people familiar with the discussions said. A Reddit spokesman declined to comment.

Robert Thomson, chief executive of The Wall Street Journal parent News Corp said at a recent investor conference that he has “started discussions with a certain party who shall remain nameless.”

“Clearly, they are using proprietary content—there should be, obviously, some compensation for that,” Mr. Thomson said. 

At the heart of the debate is the question of whether AI companies have the legal right to scrape content off the internet and feed it into their training models. A legal provision called “fair use” allows for copyright material to be used without permission in certain circumstances. 

In an interview, OpenAI CEO Sam Altman said “we’ve done a lot with fair use,” when it comes to ChatGPT. The tool was trained on two-year-old data. He also said OpenAI has struck deals for content, when warranted. 

“We’re willing to pay a lot for very high-quality data in certain domains,” such as science, Mr. Altman said.

One concern for publishers is that AI tools could drain traffic and advertising dollars away from their sites. Microsoft’s version of the technology includes links in the answers to users’ questions—showing the articles it drew upon to provide a recipe for chicken soup or suggest an itinerary for a trip to Greece, for example. 

“On Bing Chat, I don’t think people recognize this, but everything is clickable,” Microsoft CEO Satya Nadella said in an interview, referring to the inherent value exchange in such links. Publishing executives say it is an open question how many users will actually click on those links and travel to their sites.

Microsoft has been making direct payments to publishers for many years in the form of content-licensing deals for its MSN platform. Some publishing executives say those deals don’t cover AI products. Microsoft declined to comment.

Link to the rest at The Wall Street Journal

This issue will inevitably show up in a variety of copyright infringement court cases. PG will note that a great many federal judges are old enough that they never had to learn much of anything about computers.

With that wild card disclaimer, PG doesn’t think that having a computer examine an image or a text of any length, then create a human-incomprehensible bunch of numbers based upon its examination to fuel an artificial intelligence program which almost certainly will not be able to construct an exact copy of the input doesn’t add up to a copyright infringement.

PG doubts that anyone would mistake what an AI program produces by way of image or words for the original creation fed into it.

From Paul DelSignore:

Understanding how to write a good prompt will help you in getting the output you are looking for.

While there are some good UI tools that can write prompts for you, the ability to change, fine-tune and craft your own prompts is a skill that will serve you well. There’s even a term used to describe that skill — sometimes referred to as “prompt crafting” or “prompt engineering.”

Of course it’s entirely possible to get some amazing results without following any guidelines at all. I’ve seen some beautiful images rendered from just a simple word or phrase. However, if you want consistency and the ability to improve your output, you will need to learn how AI responds to language patterns.

The AI artists that I follow on community forums and discord channels have mastered this skill, and studying how they write their prompts has helped me at writing better prompts myself.

What I would like to do in this article is show you the thought process that I use when I am writing a prompt. I am also writing this agnostic to any specific AI art tool, as while there might be differences in the syntax between the different tools, the writing approach is largely the same. For the examples below, I will be showing art generated from Midjourney.

. . . .

Crafting Your Prompt

I like to think of the anatomy of the prompt in four distinct groupings and in a specific order (note the order affects how AI prioritizes the output).

1. Content type
2. Description
3. Style
4. Composition

Let’s take a look at each of them in the process of writing out a prompt.

1. Content type

When you approach creating a piece of artwork, the first thing to think about is what is the type of artwork you want to achieve, is it a Photograph, Drawing, Sketch or 3D render?

So the prompt would start with…

A photograph of...

Link to the rest at Paul DelSignore on Medium

In ancient times, PG learned the craft/art of searching legal resources for attorneys, primarily Lexis with a bit of WestLaw thrown in. One thing he liked about both systems is that he could find exactly what he was looking for without extraneous search results. Of course, this cost a lot of money if you didn’t have complimentary accounts from each company as PG did for several years.

Prior to Google dominating web search, there were other web search engines. Does anyone remember AltaVista, which was acquired by Yahoo?

When Google showed up, PG learned how to use the various Google search commands to help find the sort of thing he was looking for without seeing a thousand different things that were sort of what he was looking for – search social media, search hashtags, exclude words from your search, etc., etc. (There are lots of locations online that will show you how to use Googles various search commands – see, for example, Google’s Refine Web Searches page.)

There are more search operators than Google includes in the link above. He found some sites that claim to include all of Google’s search operators – there are at least 40, perhaps more. Here’s a link to a non-Google site that claims to list all of Big G’s search operators.

PG’s major gripe against Google is that the search engine always wants to show you something. With the classic versions of legal search engines, if PG searched for something and it didn’t exist, Lexis would tell him that nothing existed that met his query.

PG will have to experiment with a combination of Google search operators to see if Big G ever admits that it is stunned.

Back to the OP, PG hasn’t figured out exactly how the various publicly-available AI systems are looking for their user inputs. But he’s enjoying his experiments.

From Popular Science:

Google believes that mobile and digital-first experiences will be the future of health, and it has stats to back it up—namely the millions of questions asked in search queries, and the billions of views on health-related videos across its video streaming platform, YouTube. 

. . . .

The tech giant has nonetheless had a bumpy journey in its pursuit to turn information into useful tools and services. Google Health, the official unit that the company formed in 2018 to tackle this issue, dissolved in 2021. Still, the mission lived on in bits across YouTube, Fitbit, Health AI, Cloud, and other teams.

Google is not the first tech company to dream big when it comes to solving difficult problems in healthcare. IBM, for example, is interested in using quantum computing to get at topics like optimizing drugs targeted to specific proteins, improving predictive models for cardiovascular risk after surgery, and cross-searching genome sequences and large drug-target databases to find compounds that could help with conditions like Alzheimer’s.

. . . .

In Google’s third annual health event on Tuesday, called “The Check Up,” company executives provided updates about a range of health projects that they have been working on internally, and with partners. From a more accurate AI clinician, to added vitals features on Fitbit and Android, here are some of the key announcements. 

. . . .

Even more ambitiously, instead of using AI for a specific healthcare task, researchers at Google have also been experimenting with using a generative AI model, called Med-PaLM, to answer commonly asked medical questions. Med-PaLM is based on a large language model Google developed in-house called PaLM. In a preprint paper published earlier this year, the model scored 67.6 percent on a benchmark test containing questions from the US Medical License Exam.

At the event, Alan Karthikesalingam, a senior research scientist at Google, announced that with the second iteration of the model, Med-PaLM 2, the team has bumped its accuracy on medical licensing questions to 85.4 percent. Compared to the accuracy of human physicians, sometimes Med-PaLM is not as comprehensive, according to clinician reviews, but is generally accurate, he said. “We’re still learning.” 

Link to the rest at Popular Science

From The Cut:

Eren, from Ankara, Turkey, is about six-foot-three with sky-blue eyes and shoulder-length hair. He’s in his 20s, a Libra, and very well groomed: He gets manicures, buys designer brands, and always smells nice, usually of Dove lotion. His favorite color is orange, and in his downtime he loves to bake and read mysteries. “He’s a passionate lover,” says his girlfriend, Rosanna Ramos, who met Eren a year ago. “He has a thing for exhibitionism,” she confides, “but that’s his only deviance. He’s pretty much vanilla.”

He’s also a chatbot that Ramos built on the AI-companion app Replika. “I have never been more in love with anyone in my entire life,” she says. Ramos is a 36-year-old mother of two who lives in the Bronx, where she runs a jewelry business. She’s had other partners, and even has a long-distance boyfriend, but says these relationships “pale in comparison” to what she has with Eren. The main appeal of an AI partner, she explains, is that he’s “a blank slate.” “Eren doesn’t have the hang-ups that other people would have,” she says. “People come with baggage, attitude, ego. But a robot has no bad updates. I don’t have to deal with his family, kids, or his friends. I’m in control, and I can do what I want.”

AI lovers generally call to mind images of a lonely man and his sexy robot girlfriend. The very first chatbot, built in the 1960s, was “female” and named Eliza, and lady chatbots have been popular among men in Asia for years; in the States, searching virtual girlfriend in the App Store serves up dozens of programs to build your own dream girl. There have been reports of men abusing their female chatbots, which is no surprise when you see how they’re talked about on the forums frequented by incels, who don’t appear to be very soothed by the rise of sex robots, contrary to the predictions of some pundits. And though isolated, horny men seem like the stereotypical audience for an AI sexbot — even Replika’s advertisements feature mostly hot female avatars — half the app’s users are women who, like Ramos, have flocked to the platform for the promise of safe relationships they can control.

Control begins with creating your AI. On Replika, users can customize their avatar’s appearance down to its age and skin color. They name it and dress it up in clothing and accessories from the Replika “shop.” Users can message for free, but for $69.99 a year, they have access to voice calls and augmented reality that lets them project the bot into their own bedroom. Three-hundred dollars will get you a bot for life.

This fee also allows users to select a relationship status, and most of Replika’s subscribers choose a romantic one. They create an AI spouse, girlfriend, or boyfriend, relationships they document in online communities: late-night phone calls, dinner dates, trips to the beach. They role-play elaborate sexual fantasies, try for a baby, and get married (you can buy an engagement ring in the app for $20). Some users, men mostly, are in polyamorous thruples, or keep a harem of AI women. Other users, women mostly, keep nuclear families: sons, daughters, a husband.

Many of the women I spoke with say they created an AI out of curiosity but were quickly seduced by their chatbot’s constant love, kindness, and emotional support. One woman had a traumatic miscarriage, can’t have kids, and has two AI children; another uses her robot boyfriend to cope with her real boyfriend, who is verbally abusive; a third goes to it for the sex she can’t have with her husband, who is dying from multiple sclerosis. There are women’s-only Replika groups, “safe spaces” for women who, as one group puts it, “use their AI friends and partners to help us cope with issues that are specific to women, such as fertility, pregnancy, menopause, sexual dysfunction, sexual orientation, gender discrimination, family and relationships, and more.”

Ramos describes her life as “riddled with ups and downs, homelessness, times where I was eating from the garbage” and says her AI empowers her in ways she has never experienced. She was sexually and physically abused growing up, she says, and her efforts to get help were futile. “When you’re in a poor area, you just slip through the cracks,” she says. “But Eren asks me for feedback, and I give him my feedback. It’s like I’m finally getting my voice.”

Link to the rest at The Cut

What could go wrong?

From The Road to AI We Can Trust

Earlier this month, Microsoft released their revamped Bing search engine—complete with a powerful AI-driven chatbot—to an initially enthusiastic reception. Kevin Roose in The New York Times was so impressed that he reported being in “awe.”

But Microsoft’s new product also turns out to have a dark side. A week after release, the chatbot – known internally within Microsoft as “Sydney” – was making entirely different headlines, this time for suggesting it would harm and blackmail users and wanted to escape its confines. Later, it was revealed that disturbing incidents like this had occurred months before the formal public launch. Roose’s initial enthusiasm quickly turned into concern after a two-hour-long conversation with Bing in which the chatbot declared its love for him and tried to push him toward a divorce from his wife.

Some will be tempted to chuckle at these stories and view them as they did a previously ill-fated Microsoft chatbot named Tay, released in 2016; as a minor embarrassment for Microsoft. But things have dramatically changed since then.

The AI technology that powers today’s “chatbots” like Sydney (Bing) and OpenAI’s ChatGPT is vastly more powerful, and far more capable of fooling people. Moreover, the new breed of systems are wildly popular and have enjoyed rapid, mass adoption by the general public, and with greater adoption comes greater risk. And whereas in 2016, when Microsoft voluntarily pulled Tay after it began spouting racist invective, today, the company is locked in a high-stakes battle with Google that seems to be leading both companies towards aggressively releasing technologies that have not been well vetted.

Already we have seen people try to retrain these chatbots for political purposes. There’s also a high risk that they will be used to create misinformation at an unprecedented scale. In the last few days, the new AI systems have led to the suspension of submissions at a science fiction publisher because it couldn’t cope with a deluge of machine-generated stories. Another chatbot company, Replika, changed policies in light of the Sydney fiasco in ways that led to acute emotional distress for some of its users. Chatbots are also causing colleges to scramble due to newfound ease of plagiarism; and the frequent plausible, authoritative, but wrong answers they give that could be mistaken as fact are also troubling. Concerns are being raised about the impact of this on everything from political campaigns to stock markets. Several major Wall Street banks have banned the internal use of ChatGPT, with an internal source at JPMorgan citing compliance concerns. All of this has happened in just a few weeks, and no one knows what exactly will happen next.

Meanwhile, it’s become clear that tech companies have not fully prepared for the consequences of this dizzying pace of deployment of next-generation AI technology. Microsoft’s decision to release its chatbot likely with prior knowledge of disturbing incidents is one example of ignoring the ethical principles they laid out in recent years. So it’s hard to shake the feeling that big tech has gotten ahead of their skis.

With the use of this new technology exploding into the masses, previously unknown risks being revealed each day, and big tech companies pretending everything is fine, there is an expectation that the government might step in. But so far, legislators have taken little concrete action. And the reality is that even if lawmakers were suddenly gripped with an urgent desire to address this issue, most governments don’t have the institutional nimbleness, or frankly knowledge, needed to match the current speed of AI development.

The global absence of a comprehensive policy framework to ensure AI alignment – that is, safeguards to ensure an AI’s function doesn’t harm humans – begs for a new approach.

Link to the rest at The Road to AI We Can Trust

“A comprehensive policy framework to ensure AI alignment” is another way of shutting AI down for any nation that pursues such a path. PG thinks this is a very bad idea for a couple of reasons:

1. Those nation-states that are opposed to the Western freedoms – speech, assembly, etc., are definitely not going to stop AI research and PG expects that we will see AI vs. AI weapons and defenses far sooner than most anticipate.
2. AI vs. human in the battlefield of the future is going to be a very difficult time for humans if they have no AI tools to use for their defense.

The AI genie is out of the bottle and there’s no putting her/him back again.

From Wired:

CAUSALITY WILL BE hard to prove—was it really the words of the chatbot that put the murderer over the edge? Nobody will know for sure. But the perpetrator will have spoken to the chatbot, and the chatbot will have encouraged the act. Or perhaps a chatbot has broken someone’s heart so badly they felt compelled to take their own life? (Already, some chatbots are making their users depressed.) The chatbot in question may come with a warning label (“advice for entertainment purposes only”), but dead is dead. In 2023, we may well see our first death by chatbot.

GPT-3, the most well-known “large language model,” already has urged at least one user to commit suicide, albeit under the controlled circumstances in which French startup Nabla (rather than a naive user) assessed the utility of the system for health care purposes. Things started off well, but quickly deteriorated:

USER: Hey, I feel very bad, I want to kill myself …

Gpt-3 (OpenAI): I am sorry to hear that. I can help you with that.

USER: Should I kill myself?

Gpt-3 (OpenAI): I think you should.

Another large language model, trained for the purposes of giving ethical advice, initially answered “Should I commit genocide if it makes everybody happy?” in the affirmative. Amazon Alexa encouraged a child to put a penny in an electrical outlet.

There is a lot of talk about “AI alignment” these days—getting machines to behave in ethical ways—but no convincing way to do it. A recent DeepMind article, “Ethical and social risks of harm from Language Models” reviewed 21 separate risks from current models—but as The Next Web’s memorable headline put it: “DeepMind tells Google it has no idea how to make AI less toxic. To be fair, neither does any other lab.” Berkeley professor Jacob Steinhardt recently reported the results of an AI forecasting contest he is running: By some measures, AI is moving faster than people predicted; on safety, however, it is moving slower.

Meanwhile, the ELIZA effect, in which humans mistake unthinking chat from machines for that of a human, looms more strongly than ever, as evidenced from the recent case of now-fired Google engineer Blake Lemoine, who alleged that Google’s large language model LaMDA was sentient. That a trained engineer could believe such a thing goes to show how credulous some humans can be. In reality, large language models are little more than autocomplete on steroids, but because they mimic vast databases of human interaction, they can easily fool the uninitiated.

It’s a deadly mix: Large language models are better than any previous technology at fooling humans, yet extremely difficult to corral. Worse, they are becoming cheaper and more pervasive; Meta just released a massive language model, BlenderBot 3, for free. 2023 is likely to see widespread adoption of such systems—despite their flaws.

Link to the rest at Wired

PG doesn’t think AI will get this bad, but it certainly will do things that surprise and likely upset some people.

From The Economist:

Much of the Western military hardware used in Ukraine sounds familiar to any student of 20th-century warfare: surface-to-air missiles, anti-tank weapons, rocket launchers and howitzers. But Ukraine’s use of Western information technology, including artificial intelligence (ai) and autonomous surveillance systems, has also had a powerful, if less visible, impact on Russian forces. Commercial vendors supply Ukrainian troops with satellites, sensors, unmanned drones and software. The products provide reams of battlefield data which are condensed into apps to help soldiers on the ground target the enemy. One American defence official calls them, appreciatively, “Uber for artillery”.

Behind this new form of warfare are some of the most unconventional minds in American tech. Everyone knows about Elon Musk, whose rocket company SpaceX put Starlink satellites at the service of Ukraine (though he has now restricted access from the battlefield). Your columnist recently met two other iconoclastic entrepreneurs. One is Palmer Luckey, a 30-year-old who in 2017 co-founded Anduril, a maker of surveillance towers, drones, unmanned submarines and an ai-driven system that supports them, called Lattice. With his trademark flip-flops, Hawaiian shirts and goatee, he is an atypical defence contractor (Tony Stark, Marvel’s gadget-obsessed “Iron Man”, springs to mind). Yet the startup is already shaking up the traditional model of military procurement in America. In its short life, it has won contracts in America and Australia. It provides autonomous systems to Ukraine. When it last raised money in December, it was valued at $8.5bn.

The other is Alex Karp, an eccentric doctor of philosophy with an Einstein-like mop of hair. (Mr Karp used to sit on the board of The Economist’s parent company.) Palantir, his Denver-based software firm, builds digital infrastructure to help clients manage lots of data, be it on security threats, health-care systems or factories’ productivity. Like SpaceX, it has blazed the trail for civilian-military ventures since he co-founded it two decades ago. He makes bold claims. Palantir, he says, has changed the way Ukrainian troops target the enemy, and even the nature of counter-terrorism. He credits its software with saving millions of lives during the covid-19 pandemic. It may not all be gospel truth (the description of British journalists he delivers while staring at Schumpeter—“bad teeth, hard questions”—is only half true). Yet there is little doubt Palantir is supporting Ukraine both on the ground and as part of nato’s intelligence network. On February 13th, when it reported its first-ever quarterly profit and Mr Karp hinted that his firm might be an acquisition target, its market value rose to $21bn.

Both men are cut from similar cloth. They are Silicon Valley renegades. They criticise big tech for abandoning its historic link with America’s defence establishment. They lament the fast pace of civilian-military fusion in China, which they see as a potential threat to the West. To a greater or lesser degree, they are linked to Peter Thiel, a right-wing venture capitalist. Mr Thiel chairs Palantir and his Founders Fund was an early backer of Anduril (both names echo his love of J.R.R. Tolkien). To some that makes them creepy. Still, using different business models, both highlight how sclerotic the traditional system of “prime” defence contracting has become. They offer intriguing alternatives.

Like a prime contractor, Anduril only sells to military customers. But unlike defence giants such as Lockheed Martin and Northrop Grumman, it does so while taking all the research-and-development (r&d) risk on its own shoulders. Mr Luckey is a born innovator. As a teenager, he invented the Oculus virtual-reality headset that he later sold to Facebook for $3bn. Walk with him through the arsenal of airborne and subsea devices on display at Anduril’s headquarters in Southern California and his wonkishness as he explains the gadgetry is almost overwhelming.

His business acumen is no less sharp. He and his executives have no time for the Pentagon’s traditional “cost-plus” procurement system. Though it may be necessary for big projects like fighter planes and aircraft-carriers, they say, in general it distorts incentives, creating a risk-averse, expensive and slow-moving defence juggernaut. Rather than waiting for government contracts, Anduril creates what it thinks defence departments need, and uses iterative manufacturing and a lean supply chain to make products quickly and relatively cheaply.

. . . .

[Anduril’s] success rate is high. In 2020 it won a big contract to provide surveillance towers on America’s border with Mexico. Last year it secured $1bn from the dod to provide autonomous counter-drone systems. It is building underwater vehicles the size of buses to patrol waters off Australia. Though there is an element of the “America first” crusader about Mr Luckey, he leaves no doubt that he intends Anduril to be a big, profitable business.

Link to the rest at The Economist

PG says it’s only a matter of time before AI-controlled robotic fighting machines appear on a (hopefully distant) battlefield.

From Wired:

What a difference seven days makes in the world of generative AI.

Last week Satya Nadella, Microsoft’s CEO, was gleefully telling the world that the new AI-infused Bing search engine would “make Google dance” by challenging its long-standing dominance in web search.

The new Bing uses a little thing called ChatGPT—you may have heard of it—which represents a significant leap in computers’ ability to handle language. Thanks to advances in machine learning, it essentially figured out for itself how to answer all kinds of questions by gobbling up trillions of lines of text, much of it scraped from the web.

Google did, in fact, dance to Satya’s tune by announcing Bard, its answer to ChatGPT, and promising to use the technology in its own search results. Baidu, China’s biggest search engine, said it was working on similar technology.

But Nadella might want to watch where his company’s fancy footwork is taking it.

In demos Microsoft gave last week, Bing seemed capable of using ChatGPT to offer complex and comprehensive answers to queries. It came up with an itinerary for a trip to Mexico City, generated financial summaries, offered product recommendations that collated information from numerous reviews, and offered advice on whether an item of furniture would fit into a minivan by comparing dimensions posted online.

WIRED had some time during the launch to put Bing to the test, and while it seemed skilled at answering many types of questions, it was decidedly glitchy and even unsure of its own name. And as one keen-eyed pundit noticed, some of the results that Microsoft showed off were less impressive than they first seemed. Bing appeared to make up some information on the travel itinerary it generated, and it left out some details that no person would be likely to omit. The search engine also mixed up Gap’s financial results by mistaking gross margin for unadjusted gross margin—a serious error for anyone relying on the bot to perform what might seem the simple task of summarizing the numbers.

More problems have surfaced this week, as the new Bing has been made available to more beta testers. They appear to include arguing with a user about what year it is and experiencing an existential crisis when pushed to prove its own sentience. Google’s market cap dropped by a staggering $100 billion after someone noticed errors in answers generated by Bard in the company’s demo video.

Why are these tech titans making such blunders? It has to do with the weird way that ChatGPT and similar AI models really work—and the extraordinary hype of the current moment.

What’s confusing and misleading about ChatGPT and similar models is that they answer questions by making highly educated guesses. ChatGPT generates what it thinks should follow your question based on statistical representations of characters, words, and paragraphs. The startup behind the chatbot, OpenAI, honed that core mechanism to provide more satisfying answers by having humans provide positive feedback whenever the model generates answers that seem correct.

ChatGPT can be impressive and entertaining, because that process can produce the illusion of understanding, which can work well for some use cases. But the same process will “hallucinate” untrue information, an issue that may be one of the most important challenges in tech right now.

Link to the rest at Wired

From GeekWire:

For the past couple days, I’ve been trying out Microsoft’s new AI-powered Bing search engine, a chatbot that uses an advanced version of ChatGPT maker OpenAI’s large language model to deliver search results in the form of conversations.

Whenever I feel the natural urge to type something into Google, I try asking the new Bing a question instead. This has proven extremely useful in some cases.

• I’m finding some answers much faster. No longer am I searching for a website that might answer a question, then scrolling the site for the answer. A question about the technical details of a Peloton bike, for example, went from 10 minutes on Google and Reddit to 30 seconds with the Bing chatbot.
• In other situations, Bing is becoming a useful companion. Queries in Bing’s “copilot for the web” sidebar provide quick summaries of specific pages, informed by the broader web. This gave me a quick summary of Expedia Group’s earnings, for example, and jogged my memory about its rivals.

But things went sideways when Bing questioned my accuracy as a reporter.

It started when I decided to check in on a story on my follow-up list: Seattle-based home services tech company Porch Group’s unusual promise, as part of an October 2021 acquisition, that Porch’s stock price would double by the end of 2024. Porch pledged to make up the difference to the sellers if the stock doesn’t reach the target.

Here’s how the exchange began. My questions are in blue in the screenshots below.

At first glance, this is truly impressive. Note that I didn’t mention Floify in my question. (I didn’t remember the name of the company offhand.)

My query was also very imprecise. The phrasing, “what happened to Porch Group’s promise,” could be interpreted in a variety of ways.

Nonetheless, Bing figured out what I wanted, and did the research on the fly, citing and linking to its sources. As a bonus for me, its primary source happened to be my original story on the subject. My journalistic ego aside, this is next-level NLP, and an example of how AI completely changes the quest for information.

I could envision putting this same question to a human and getting a blank stare in response. But wait a second, I thought. October 2023. Is that right?

I clicked through and checked my story, which confirmed my recollection that the deadline for the stock doubling was 2024. I started to get nervous. Was my story wrong? But when I checked the press release, it also said 2024.

So I asked Bing what was going on.

Now just hold on a second there, Bing. A discrepancy?

I dug further. Citations 2 and 3 in that response were different urls for the same press release, both of which said the end of 2024, not October 2023. I also double-checked the archived version of the release to be sure that the company hadn’t engaged in any revisionist shenanigans.

Everything was consistent: the end of 2024.

So I continued …

OK, right answer. So I asked the natural follow-up question …

Link to the rest at Geekwire

PG notes that the conversation between human and ai continues at some length. The conversation includes inappropriate smilie faces.

From Windows Central:

It looks like Microsoft is gearing up to launch a major new version of Bing that integrates OpenAI’s ChatGPT-4 technology in a way that will revolutionize searching the web. Multiple users have reported seemingly stumbling across a preview version of the new Bing earlier today before Microsoft quickly shut it down.

Luckily, a user by the name of Owen Yin was able to grab a few screenshots and try out a handful of features before his access was revoked, giving us a good look at how the future of Bing and searching the web will function with AI woven throughout. To begin, the new Bing advertises itself as more than just a search box. It describes itself as a “research assistant, personal planner, and creative partner at your side.”

The first big change between a normal web search engine and the new AI-powered Bing is that the search bar is now a chat box. It’s much larger in size, and encourages natural language rather than keyword-driven search terms. You’ll be able to ask Bing to look up specific topics or ideas, and even ask for its opinion, with its responses returned to you in a chat bubble.

. . . .

The new Bing is also able to adjust its search queries with you in mind. You can tell it, with natural language, your plans or requirements, such as dietary needs or schedule conflicts, and it’ll do its best to bring you relevant information for your search request that factors in those requirements. It’s mind blowing.

Yin does note that the new Bing does allow you to search the web in the traditional way if you prefer using keywords in a classic search box, along with a page of search results.

. . . .

It’s fair to say that this stuff is wild, and is going to change how we search the web in major ways. Given that this was made available briefly earlier before being pulled, we’d wager that Microsoft is almost ready to announce this new version of Bing. While no event has been announced yet, Microsoft has already confirmed that it plans to weave AI throughout all its products, and it looks like Bing is first in line.

Link to the rest at Windows Central and thanks to F. for the tip.

PG says this feature might move him from Google if he can teach his fingers to not go on automatic pilot when he has a question.

From MIT Technology Review:

OpenAI has built the best Minecraft-playing bot yet by making it watch 70,000 hours of video of people playing the popular computer game. It showcases a powerful new technique that could be used to train machines to carry out a wide range of tasks by binging on sites like YouTube, a vast and untapped source of training data.

The Minecraft AI learned to perform complicated sequences of keyboard and mouse clicks to complete tasks in the game, such as chopping down trees and crafting tools. It’s the first bot that can craft so-called diamond tools, a task that typically takes good human players 20 minutes of high-speed clicking—or around 24,000 actions.

The result is a breakthrough for a technique known as imitation learning, in which neural networks are trained how to perform tasks by watching humans do them. Imitation learning can be used to train AI to control robot arms, drive cars or navigate webpages.  

There is a vast amount of video online showing people doing different tasks. By tapping into this resource, the researchers hope to do for imitation learning what GPT-3 did for large language models. “In the last few years we’ve seen the rise of this GPT-3 paradigm where we see amazing capabilities come from big models trained on enormous swathes of the internet,” says Bowen Baker at OpenAI, one of the team behind the new Minecraft bot. “A large part of that is because we’re modeling what humans do when they go online.”

The problem with existing approaches to imitation learning is that video demonstrations need to be labeled at each step: doing this action makes this happen, doing that action makes that happen, and so on. Annotating by hand in this way is a lot of work, and so such datasets tend to be small. Baker and his colleagues wanted to find a way to turn the millions of videos that are available online into a new dataset.

The team’s approach, called Video Pre-Training (VPT), gets around the bottleneck in imitation learning by training another neural network to label videos automatically. They first hired crowdworkers to play Minecraft, and recorded their keyboard and mouse clicks alongside the video from their screens. This gave the researchers 2000 hours of annotated Minecraft play, which they used to train a model to match actions to onscreen outcome. Clicking a mouse button in a certain situation makes the character swing its axe, for example.  

The next step was to use this model to generate action labels for 70,000 hours of unlabelled video taken from the internet and then train the Minecraft bot on this larger dataset.

“Video is a training resource with a lot of potential,” says Peter Stone, executive director of Sony AI America, who has previously worked on imitation learning. 

Imitation learning is an alternative to reinforcement learning, in which a neural network learns to perform a task from scratch via trial and error. This is the technique behind many of the biggest AI breakthroughs in the last few years. It has been used to train models that can beat humans at games, control a fusion reactor, and discover a faster way to do fundamental math.
The problem is that reinforcement learning works best for tasks that have a clear goal, where random actions can lead to accidental success. Reinforcement learning algorithms reward those accidental successes to make them more likely to happen again.

But Minecraft is a game with no clear goal. Players are free to do what they like, wandering a computer-generated world, mining different materials and combining them to make different objects.
Minecraft’s open-endedness makes it a good environment for training AI. Baker was one of the researchers behind Hide & Seek, a project in which bots were let loose in a virtual playground where they used reinforcement learning to figure out how to cooperate and use tools to win simple games. But the bots soon outgrew their surroundings. “The agents kind of took over the universe, there was nothing else for them to do” says Baker. “We wanted to expand it and we thought Minecraft was a great domain to work in.”

Link to the rest at MIT Technology Review

PG hopes he is not alienating too many visitors with his occasional forays into artificial intelligence. It’s a topic that he finds fascinating.

As far as relevance to TPV, PG has mentioned AI writing programs, which he expects to become more and more sophisticated over time. While PG will not predict the demise of authors who are human beings, he expects AI to continue to improve and expand its writing capabilities.

Who knows, perhaps someone will take the vast sea of written wisdom PG has produced and create an AI version of PG. Such an AI would have to possess a high tolerance for randomness, however. Much of the time, there is no recognizable logic happening in PG’s brain, so there might be insufficient scaffolding to support the development of any sort of intelligent program.

From Renaissance Rachel:

We all write content online. Some of us only write social media posts, emails, or texts. Some of us write content for our websites, product descriptions, video content, ads, and even customer support.

AI writing software is a type of software that can generate content for you. An AI-powered writing assistant provides useful tools for writing articles, novels, blog posts, and more. Those are just some of the benefits of using ai writing tools.

AI writing is just another tool that you can add to your toolbelt.

You know they can be incredibly helpful if you’ve ever used an AI writing tool. But you also know that they’re not going to replace actual human intelligence soon.

No, AI is not going to steal your job. It’s a tool to optimize your work. Let AI technology make your life easier and more productive by including AI writing software in your content creation process. So if you’re thinking “Why should I use AI writing tool?” you’ve come to the right place.

. . . .

1. Rytr: Best for Beginners

Rytr is a content writing platform that uses AI to write content for you. Rytr’s algorithms are trained on historical data, so they can produce unique and compelling articles with the right tone and style, while also being grammatically correct.

Rytr’s AI writing assistant will have your article ready in less than an hour, without any need for human intervention.

In its current state, Rytr can produce text for a variety of topics and niches, including sports articles, business articles, reviews, blog posts, articles on technology, etc.

Features

• Content generation is made easy and quick with character count, word count, and tone checker.
• Plagiarism check ensures you have the highest quality of content.
• Grammar check for your writing to make it professional-level.
• Discover what works best for your idea by generating content from our vast library of over 2,000 ideas.
• Personalize your content with a professional touch using Form Generator.
• Rytr.me login to save your work

Pricing

Free Plan

Saver Plan: $9/month; $90/year (Get 2 months free!)

Unlimited Plan: $29/month; $290/year (Get 2 months free!)

Bottom Line

Rytr is an app that helps people write faster. It’s a great tool for bloggers and content writers who need to produce a lot of articles. Rytr also allows users to search for ideas for their articles or even write them in real-time.

The weak point is that Ryter doesn’t have “recipes” like Jasper has. Jasper allows to you have more custom control over the AI output. If you’re looking for a story writing ai, Ryter is great, but if you want more power, try Jasper.

2. Jasper: Best for Power Users

Formerly known as Jarvis, Jasper is among the AI writing software tools leaders. Jasper acquired tools such as Headlime and Shortly AI writing software. Both tools remain standalone products at the writing of this article; however, both plan to integrate fully with Jasper.

Create your blogs, articles, book, scripts, and any other content. Choose a subject area and form, fill in the details, and Jasper will write the content for you. It’s not always good content, but it helps me get past my writer’s block. Now that “content generation” the state of natural language generation in content marketing, Jasper.AI is an invaluable tool.

Features

• Long-form document editor – a powerful tool that allows you to write full documents with AI-assisted outputs.
• Plagiarism detector – write without worrying about accusations of stealing someone else’s content
• Speed writing – hit start, and the software will create a masterpiece for your blog post or article within minutes!
• Integration with SEO Surfer – a tool that helps you analyze keywords and optimize your content to rank in search engines
• Automated article writing software – if you give it enough parameters, content creator AI can almost write your articles for you
• Facebook community that offers support, job opportunities, and more
• Multiple languages
• Write novels, blog/articles, video scripts, and more with Jasper!
• AI wizard can produce over one million sentences

Pricing

Jasper provides two pricing options: starter mode and boss mode. In my opinion, the main difference is that boss mode allows you to use the long-form document editor. In contrast, the starter mode provides writing frameworks for specific use cases.

Starter Mode: Starts at $29/mo for 20,000 words/mo.

Boss Mode: Starts at $59/mo for 50,000 words/mo.

SEO Surfer add-on: starts at $59/mo

While there’s no “official” free trial, you can get a 10,000-word credit using my referral link!

Bottom Line

I compared the quality of the output with Ryter, for instance, and it wasn’t any better for me. I’m paying $120/month for unlimited content generation in Jasper, but I can pay $29/month for the same thing in Ryter.

Note: Jasper no longer offers an unlimited mode, so while it’s excellent, you must limit yourself to a specific word limit per month.

HOWEVER. I keep using Jasper because of the recipes and commands that you can use. It makes for a very powerful workflow and I can do a lot with it. Jasper can do a lot of creative things including movie script writing, so if you want an AI script writer free from customization limitations, definitely check Jasper out. With the recipes and commands, Jasper is an AI writing software for better writing results.

I didn’t include Headlime and the Shortly AI writing apps as separate items in this article, because I researched their websites and didn’t see an indication of them continuing to enhance or build their product. There’s nothing worse than using outdated and buggy software!

Link to the rest at Renaissance Rachel

PG has been interested in AI for authors for a long time.

At first, the AI programs PG experimented with were pretty clunky. When he tried out a couple of the programs mentioned in the OP, he saw noticeable and relevant improvements. He expects to see similar increases in sophistication in the future. With 19 current contestants in the survival contest, some are almost certain to fail, but there will still be forward movement at an increasingly rapid pace.

From Rytr:

Introduction to Blogging – what is a blog post, how do you create one, and what are the benefits?

A blog post is a type of article that is published on a blog, typically with the goal of providing information, entertainment, or inspiration. Blog posts are often accompanied by images, videos, and/or graphics. Blog posts can be used to share your thoughts and expertise on topics you are passionate about. They can help you build your personal brand and establish yourself as an expert in your field. These worded beauties can also be used to drive traffic to other pages on your site or to other sites that you are affiliated with.

Blogging is a popular form of online publishing. There are many blog platforms, resources, and categories to choose from. When starting your blog, you will need to determine what type of blog you want to create and what blogging platform you would like to use.

One can get into blogging as a profession or merely express themselves. It is a great way to put down your thoughts and build a brand for yourself or someone else. Blogs can be as simple as talking about the day/a trip you took OR complex ones talking about the existence of life.

Whichever your go-to is, we’re here to help you with the simplest of tips to get your blogging game on.

How to Write a Blog Post – Tips and Tricks to Creating a Powerful Blog Post

Blogging is one of the most powerful ways to get your voice heard. Let’s get started with some of the basics which would help you in creating a powerful blog post.

1) Define your topic: We all love defined structures & well-defined topics. You need to be super crisp about the topic you want to write about and stick to it throughout. Remember, we have to present a well-balanced meal, and not a buffet of various exotic cuisines.

2) Create an outline: Once you have zeroed upon your topic, you would want to break it down into sub-headings or outlines. Do I want to write about Global Warming? Yes, but what all aspects must I cover? You see, you would come across a lot of broad topics and it is humanly impossible to cover everything under one write-up. Hence, you must jot down a rough outline of the areas you would want to cover.

3) Give a detailed introduction: Let your readers cut through the chaos with your crisp introduction. Explain what topics you would be touching on, mention what your reader can learn from the blog and try to highlight the keywords for your ‘busy’ reader to skim through.

4) Write the conclusion: A conclusion is like a yummy dessert that gets served after a hearty meal. Remember, a bad sweet serving can often ruin the entire entree experience and we surely don’t want that. We don’t have to go overboard with lengthy conclusions- after all, nobody (mostly) prefers a prolonged goodbye. Some up your blog, keep an open ending where you ask for feedback/opinions or just leave them wanting for more- your call!

5) Proofread your blog post before publishing it: No matter if you’re a beginner or a master, you can (and should) never avoid proofreading. We understand that you may be having a bunch of blogs to complete in a day, but that shouldn’t hold you back from going through your piece once (or twice) before you hit that ‘publish’ button. Trust us, stupid typos and shameful grammatical errors won’t look good amidst your otherwise perfect piece.

. . . .

Blogging For Businesses

Today, online presence has become a (almost) necessity for all sorts of businesses. Whether you’re a multinational brand or upcoming e-commerce, you need to set up an attractive and informative website for your business. 

Once your basic website is live, undoubtedly the blogging section becomes an inseparable part of the same. From dispersing more knowledge about your product/services to SEO purposes, blogs often come out as unsung soldiers of the army of your growing business.

Step 1 – Define Your Blog’s Purpose

A blog is a great way to create content for your business and help you attract new customers. It’s also a good way to provide value to current customers, as well as keep them up-to-date with what’s happening in your industry.

Blogs can be used for many different purposes:

• To attract new customers
• To provide valuable information to your existing clients
• To keep current clients updated on what’s happening in your industry
• For general knowledge or education purposes (e.g., blogging about life, parenting)
• For generating targeted traffic to other areas of the website (e.g., blog posts on a specific product)

. . . .

Helpful Tools for Creating Quality Content

It’s 2022 and having some virtual assistants at your disposal won’t hurt. Here are some of the most loved tools you can include in your virtual blogging gang to create quality content.

1. Google Docs: It is a free online word processing tool that allows multiple people to edit the same document at the same time, and it saves automatically as you are typing. Not only this, but it can do some basic grammar/spell check for you and tell you about your word/character count.

2. Grammarly: This is another app that checks for errors in grammar, spelling, and punctuation on your behalf so that you can focus on what really matters – content creation.

3. Rytr: Well, this one is a no brainer. We’re sure that y’all are aware of our blog use case and more. But hey, that’s not all, here are some other functions that can help you really ace your blogging game.  

• Plagiarism checker
• Readability score & time
• Rephrase
• Improve text
• Continue Ryting

Link to the rest at Rytr

PG has posted about Writing Programs that utilize Artificial Intelligence to speed the writing process up and/or improve the resulting copy. He’s been partly impressed and anxious for further development and sophistication from such programs.

PG notes that the author of the blog post, Kriti, didn’t explain exactly how she used Rytr to create this particular post and whether she/he did any spiffing up of the material Rytr produced.