Resistance is Futile

by

From studiomcah:

Yes, it’s true: after much gnashing of teeth and a token resistance to the inevitable, I decided it was time to do serious experimentation with AI, especially after hearing multiple reports, all good, about Anthropic’s Claude. To be clear, I continue to think the legal repercussions of the training of AI models on unlicensed intellectual property (whether that’s visual art, fiction or nonfiction, music, etc) need to be hashed out… and we need to decide now who owns a person’s voice, face, and personality to protect against the use of deepfakes to defame people or defraud their loved ones.

            None of that, however, changes that Artificial General Intelligence (AGI) and LLMs (Large Language Models) are not going anywhere, and are already changing things. I would rather not be drowned by the tidal wave of revolution, so for once I’m trying to surf the initial waves. “What can AI possibly do for me, if I don’t want it to write my books or draw my pictures?” I wondered, and my loved ones said, “Why don’t you find out instead of guessing?” (I am surrounded by smart people.)

            This ended up being a perfect time to experiment because (by accident!) I had a problem that needed solving: I want to set up a sales website so people can shop from me directly instead of buying from Etsy or Bandcamp or Amazon. I’d just read a book that broke down the tiers of products you want to offer, from freebies to lure in new readers, all the way to premium purchases that will only be attractive to superfans. Since my book catalog alone is over 70 titles, brainstorming what things to put in what categories sounded less like fun and more like shoveling the Augean stables. I had just signed up for Claude, so I figured: why not see if it can figure these things out for me?

            Its initial suggestions were generic based on the information I gave it—that I was an author, of 70 books, mostly science fiction, but some fantasy, children’s, romance, and nonfiction. I was also a painter. I was intrigued by the fact that it knew that ebooks made good low tier products based on price, and that premium offerings should involve autographs, special editions, or bundling with themed art or merchandise… but it was too non-specific for me.

            Which is when I fell down the rabbit hole. I discovered I could feed it my list of published works. Then my book catalog with all the covers and descriptions and tags. I gave it all-time sales data from my retailers… and then bandcamp… and then etsy… and then all my kickstarter data. I even gave it website traffic information, patreon and locals stats, and social media follower counts. With every file I fed it, I asked it to refine its ideas on how I should be positioning, bundling, and marketing my products. I asked it what underperforming books might be promising if presented to some new audience. I even asked it to find recurring themes across all my books and use that information to create marketing copy for new readers.

            Every so often I’d stop to ask it ancillary business questions, like “My large backlist can be intimidating to new readers. How do I attract them despite that?” or “I write in diverse genres, which makes my work difficult to market. How can my broad writing range be used as a strength, instead of a weakness, and how can I make new readers interested in all my offerings?” And it continued to give me sensible ideas, many of which I had already thought of, along with a few I hadn’t.

            Already I had to stop and marvel at how bizarre it was that a computer was just spouting off all this stuff in response to questions. Where does it get these answers? How does it construct them? How does it know what words mean?? It is completely inscrutable, but the interaction feels so normal that you keep going. So I did.

            By the end of that conversation, Claude knew not only which of my books and settings were bestsellers, it gave me excellent guesses on which of my themes or tropes were doing best in the market, and had used that information to craft a set of offerings for my (as yet unrealized) shop that would not only attract people with the tried-and-true series, like Dreamhealers and Her Instruments, but also tempt people with the promising but underselling ones, like Thief of Songs. “Narrow that down to ten initial offerings,” I told it, “because I want to launch my store with a limited number of items to get my feet wet.” Which it did, and they were all reasonable ideas. And I went to sleep (or tried), feeling like I’d completely underestimated the utility of LLMs. I had started the day with a tedious task I hadn’t wanted to do that required knowledge of my entire product catalog and how my art and writing interacted over the 25+ years I’d been making things, and Claude had learned enough to do it for me.

Link to the rest at studiomcah

Studiomicah is the home of author M.C.A. Hogarth, whom PG met a very long time ago. (Maggie was probably a child at the time.)

She writes science fiction, fantasy, and anthropomorphic animal genres and struck PG as a very nice woman. He expects she hasn’t changed.

Here’s a link to Maggie’s books on Amazon. If you like to read books written by nice people, check out Maggie’s work. If you would like to read books by nice people who are talented authors, Maggie will deliver you a twofer.

Click to Tweet/Email/Share This Post

28 thoughts on “Resistance is Futile”

  1. It seems the OP Borg allusion is appropriate.

    From Business Insider via Yahoo:
    https://ca.finance.yahoo.com/news/microsoft-stock-surge-another-18-224945921.html

    “Wedbush Securities hiked Microsoft’s price target from $475 to $500, touting the company’s artificial intelligence ventures as a major growth accelerant. That implies around 18% upside from the firm’s current share price.

    Leading this bullishness is the success of Microsoft’s Copilot, an AI assistant tool that Wedbush previously considered grounds for an “iPhone moment.”

    “In a nutshell, we see an acceleration of adoption for generative AI and Copilot activity which in turn is catalyzing more Azure cloud deal flow for Nadella & Co. with major momentum heading into the next 6 to 12 months as AI use cases explode across the enterprise landscape,” analyst Dan Ives wrote in a note on Tuesday, citing customer checks.

    Over the next three years, 70% of Microsoft’s base is forecast to be using the firm’s AI-driven functionalities; that’s up from Ives’ prior 60% estimate.

    As use cases roll out over the next three to six months, fiscal year 2025 should mark an inflection point for AI growth, he said. By that year, strong Copilot uptake could ultimately add between $25 billion to $30 billion to Microsoft’s revenue.”

    Also:

    “And still, all this is just the start of a new gear of growth for Microsoft, Wedbush cited, with AI leading the company’s trajectory for the coming years.

    “We continue to believe this is a ‘1995 Moment’ as this AI Revolution plays out across the tech ecosystem over the coming years with Nvidia/Jensen and Microsoft leading the way as this $1 trillion AI tidal wave hits the shores of the tech world,” Ives concluded.”

    —-

    Note the comment tbat use case discovery is just starting.
    Just as the evolution of PCs into “AI PCs” with Neural processor chips is just starting.

    The 1995 reference to the dot-com bubble is appropriate since annual US productivity growth doubled from 1.4% over 1990 to 2.8% from 1995 to 1999. Even after the dot com bust the new normal was 2.3% a year. Over the entire decade, US productivity grew 23.3%.

    That is what we’re looking at.
    Resistance *is* futile. (It took two decades but Walmart and Target and most surviving retailers adapted to rely on the web. Those that didn’t are mostly gone with the pandemic.)

    Just as with the internet, the way forward is look for use cases that help *your* workflow.

  2. I have discovered another useful use case with Claude: I feed it worldbuilding decisions I’ve made and ask it to do the tedious calculatey bits for me, like population changes over time, how big this city need to be in square miles, and basically stuff I’d need to cobble together tons of internet calculators to do instead, I can make it a much more complex calculator. That is useful. I’m inordinately fond of realism despite my very strongly nonnegotiable constraints.

    • Is its graphic image capability suitable for world map building?
      BTW, it might be worth pointing to the Claude wikipedia entry:

      https://en.m.wikipedia.org/wiki/Claude_(language_model)

      …and point out it shares roots with GPT since its home, ANTHROPIC, was founded by OpenAI founders. It’s received $7B+ in investment funding including up to $4B from Amazon and as a result it is tied to AWS like Gpt Is tied to AZURE.

      Good pedigree.

  3. Ah, I was just thinking about MCA’s books! I’m glad she’s embracing the Light Side of “AI,” and I like the use she found for it. As one of her readers, I can confirm the AIs assessment of themes and story elements I enjoy from her**. I’m going to link this post to some writing groups I know.

    Side note, in day-to-day life “Claude” looks like it’s at least doing the Virtual Intelligence assistant work I would want in an AI. Basically, AVINA from the Mass Effect games. Or the computer from Star Trek. Too bad Majel Barret*** isn’t around to be the avatar for this type of service. Anyway, I think people will more willingly embrace that aspect of AI than the kind where they threaten IPs and copyrights.

    **My way of evaluating a source/resource is to ask a question I already know the answer to. If the would-be source nails that answer, I will trust it when it answers questions I don’t know the answer to.

    ***Tricia Helfer would likely not be as effective as a spokes-avatar if the goal is to get people to trust the AI 😀

    • Too many people are enthralled by the media obfuscation over “AI” to actually consider what tge tech is, how it works and what it really does.
      Even before ChatGPT hit the mainstream, FORBES did a piece highlighting 16 industries where “AI” tools can offer masive benefits.
      Not a one has anything to do with copyrights.

      https://www.forbes.com/sites/forbestechcouncil/2022/01/13/16-industries-and-functions-that-will-benefit-from-ai-in-2022-and-beyond/?sh=55e08b32477d

      As a point of fact, MIT and Harvard (two of the more problematically woke universities) did studies on real world use of LLM and other Transformer tech and found demonstrable productivity boosts of 40-70%. That kind of value is not going to go away because the media is afraid LLMs will hurt their cozy little domain. Remember that for all their carping over the evil web, the internet is everywhere. They’re fighting the tide of time.

      As for a good voice to front AI interfaces, have you listened to Tori Holub? Her natural speaking voice strikes me as a younger, softer version of Majel Barret. Plus she can really sing. At 20 years old, she’ll be around for a while. (Look for her CLOSE TO YOU video on youtube. And her other covers show amazing vocal control.) She could probably replicate Barret with minimal effort.

      • In the Fifties, it was very common to see computers referred to as Electronic Brains and Thinking Machines. Programming a register sunk those ideas. Today, “AI” is being used the same way. But I think the LLMs have revealed a potential few had attributed to the register – or maybe mitochondria?

        • Maybe midichlorians. 😉

          My main reaction when I see somebody attribute actual sentience to software is to reject *their* intelligence. Occam’s Razor. 😀

        • Ohh, nice flash of insight now on the rules about computers in the Dune universe. I suppose the rules about “thinking machines” would have resonated with people in the 50s.

      • Nice find on Tori Holub! I like her explainer video on the Carpenters, and she has a nice songbird singing voice like Karen Carpenter 🙂

        I’m with Catherine Asaro on using this tech for research: “Computer, give me a schematic for a house with a subterranean light well. Computer, make me a map in the style of the Peutinger Table Map. Computer, talk to me about Eleanora, Judikessa of Arborea.” Except more reliable than the current iterations that Microsoft is using with Bing. This phase is the shakedown cruise, I’m sure we’ll get there soon enough.

        • Hopefully, yes.
          The sad thing is the media luddites are scaring people to protect their empires and creating roadblocks and useless narratives like this:

          https://arstechnica.com/information-technology/2024/03/worlds-first-global-ai-resolution-unanimously-adopted-by-united-nations/

          How does any of this help?

          The **United Nations General Assembly** recently adopted a **landmark resolution** on the promotion of **“safe, secure, and trustworthy” artificial intelligence (AI) systems**. Here are the key points:

          1. **Objective**: The resolution aims to bridge the AI and digital divides between and within countries and promote safe, secure, and trustworthy AI systems to accelerate progress towards the **2030 Agenda for Sustainable Development**.

          2. **Human Rights**: It emphasizes the respect, protection, and promotion of **human rights** in the design, development, deployment, and use of AI. The same rights that people have offline must also be protected online.

          3. **Global Consensus**: The resolution was co-sponsored by more than **120 Member States** and represents the first time the General Assembly has adopted a resolution on regulating AI.

          4. **Potential**: AI systems have the potential to accelerate progress towards the **17 Sustainable Development Goals** (SDGs). They are already being used for disease detection, disaster preparedness, and scientific advancements.

          5. **Responsibility**: The international community must govern AI technology to prevent misuse and ensure it benefits humanity. The resolution encourages responsible and inclusive approaches to AI.

          6. **Digital Divide**: Recognizing varying levels of technological development, the resolution urges cooperation to close the digital divide and increase digital literacy.

          Source: Conversation with Bing, 3/27/2024

          Nothing about it being useful, just excuses for luddites to strangle the tech with regulation. Which is why the developers and adopters are rushing to get it everywhere faster than the idiots can spin out the red tape. Hopefully they can move fast enough to make it a fait accompli.

          The potential is vast, but the idiots are legion.
          I remain hopeful–I have several uses in mind–but I see trouble ahead.

  4. In case anyone else has this use case, it will convert between fantasy calendars/clocks and earth ones if you give it the mapping. Between that and spreadsheeting for bulk import, I find the tool potentially useful, esp. since it’s TOS are excellent.

  5. “None of that, however, changes that Artificial General Intelligence (AGI) and LLMs (Large Language Models) are not going anywhere, and are already changing things. ”

    That’s where I checked out. Artificial General Intelligence does not exist. Furthermore, there is no apparent roadmap to creating it. That there is hype from a combination of fantasists and grifters (not entirely distinct groups) does not change this.

    • No, true AGI, is not coming anytime soon.
      Just like there is no AI in “AI”.
      But…

      The tech is real and evolving the ability to reasonably generalize from the specific to the general and use that to generate outputs. *That* is what the real world “Generalized Artificial Intelligence” software can do. (Existing models can solve math problems because they have internalized the rules of algebra, for one.)

      The buzzwords are inappropriate but the tech is viable and useful, as Ms Hograth is reporting. Don’t let the “new and improved” hype obscure the new range of useful tools. I dislike the terms myself but that is what the industry has settled on and I’m not going to let it blind me to the changes coming. I’ll adapt and adopt what suits me; there are uses to this, both small and large, but like any tool you need to be careful to separate reality from marketing.

      Above all, this is a very real tech and like all new tech it will go through phases: first, overestimation, then the disappointment it isn’t really magic, then gradual infiltration everywhere until one day it is ubiquitous. Like the internet, which went from “the information superhighway” Gore expected would magically get him elected, to an everday feature of life given no more thought than indoor plumbing.

      I ignore the buzzwords and hype but I’m seeing valuable use cases emerging…

      Other than the silly idea of AI assembled tsunami of dreck. And even that “story generation” has its uses. Remember the recent post about MICROSOFT READING COACH software that assembles topical material to teach kids to read?

      https://www.thepassivevoice.com/microsoft-reading-coach-is-now-available-as-a-stand-alone-app-for-schools/

    • Artificial General Intelligence does not exist.

      Does Real General Intelligence exist? If so, how would we distinguish the product of Real General intelligence from the product of something artificial?

      • Agency.
        Even IdiotPoliticians come up with stupid notions (“RICO is not a crime”) on their own. No prompt engineering required.

        • And how do we judge agency? Most of the discussions of human vs non-human intelligence fail to tell us what intelligence is and how it works. If human and non-human entities produce the same or similar output, might it be time to question the pedestal on which we place human intelligence?

          • Agency is acting without prodding.
            Agency is internaly driven.
            The day software can activiate and do something because it wants to, without any prompt, is the day you can *begin* to ask if it is intelligence.

            Don’t like “agency” because of its political misuse?
            Fine, use initiative. Or will to act. Self-starter.
            Whatever you call it, cats have it. They’ll demand petting when they want it. Remember the expression “herding cats”.
            “AI” Software has no more agency, will, initiative than a spreadsheet. And in fact, the core tech behind the illusion of intelligence runs fine in a spreadsheet.

            https://www.windowscentral.com/software-apps/can-an-excel-spreadsheet-with-gpt-2-replace-copilot-pro-no-but-it-shows-us-how-ai-works

            It is okay to wonder if humans are intelligent since so much of what they do kneejerking and unthinking, promoted by manipulation or delusion (especially politicians and useful idiots) but even the most clueless of activists just above the level of an earthworm, has the option to choose and do nothing. They don’t avail themselves of this power very often but they still have it.

            Software has no such power, to decide on its own to act or not. It only does or not what it is made to do. You ever read Asimov’s ROBBIE from 1941?
            First of his robot stories. It tells tbe story of an early robot taken home by a corporate insider to be a toy;companion to his daughter. And like all of his robot stories it deals with the subject of what we now think of as artificial intelligence. And in all those stories it comes down to the last sentence of the first story. “He was made that way.”

            The illusion of intelligence is not the real thing. Which the idiot google employee failed to consider. It is a fine line and one worth remembering as we move forward from humans writing software to software writing software (now) to self-updating software. The prompt may be hidden layers deep, but like a Rube Goldberg contraption, the driving will still comes from a human, not a machine.

            Asimov’s robot stories are amusing and, in the latter volumes thought provoking, but the underlying question is always the same. Robots do what they’re told to do by fallible, careless humans. That and no more. Plus the law of unintended consequences because without it, he wouldn’t have a story worth reading. And they are most definitely worth reading.

            In fact, I, ROBOT shoud be required reading in the age of “AI” and humanoid robots we are now entering.

            • Sure. Let that day dawn.

              So, how would we distinguish the product of Real General intelligence from the product of something artificial if both sources are unprompted?

              Does a human answering the above question demonstrate intelligence? The question itself is a prod.

              No test could be conclusive if a prod eliminates intelligence as a response because it was prodded. The test would be a prod.

              And just for fun, is it possible for a human to act without a prod? Can we act internally without experience (prod)? Without experience, language, prior interaction, senses? What evidence do we have that thoughts come to us unprodded?

              And Asimov cheated. The Positronic Brain was an unexplained fluke from a kid messing around with a machine. Giskard Robiwiki is the really interesting robot.

              • I never said humans (as a class) are inteligent. It’s not a requirement for self-prompting. As Forest Gump wisely said, “stupid is as stupid does”. Darwin Awards are an eternal exemplar of the extent of human stupidity.

                And yes they can be externally prompted. But they don’t *have* to be. Neither do cats.
                Or, look at babies.

                Here are some common **unprompted baby actions**:

                1. **Smiling**: Babies often smile spontaneously, even when there’s no apparent reason. It’s their way of connecting with the world around them.

                2. **Cooing and Babbling**: Babies start making sounds early on. They coo, babble, and experiment with different vocalizations. It’s their way of practicing communication.

                3. **Grabbing Objects**: As their motor skills develop, babies reach out and grab objects. They’re curious about textures and shapes.

                4. **Startle Reflex**: When babies hear a sudden noise or experience a quick movement, they might throw their arms out and then pull them back in. This is the **Moro reflex**.

                5. **Rooting Reflex**: Babies turn their heads toward anything that touches their cheek or mouth. It’s an instinctual response to find the breast for feeding.

                6. **Kicking and Wiggling**: Babies love to kick their legs and wiggle their bodies. It’s a way to explore their newfound mobility.

                7. **Sucking**: Babies have a strong sucking reflex. It helps them feed and provides comfort.

                8. **Eye Contact**: Babies seek eye contact with caregivers. It’s a crucial part of bonding and social development.

                9. **Exploring Faces**: Babies study faces intently. They’re fascinated by features like eyes, noses, and mouths.

                10. **Startling Awake**: Sometimes babies suddenly wake up from sleep, seemingly without any reason. It’s normal and often related to their sleep cycles.

                There *might* someday be a sentient software entity but I seriously doubt it. Mechanical turks the lot of ’em.

                They make for interesting and useful story elements. And yes, they’re all cheats. So is FTL. PSI abilities. Time travel. All unlikely but not known to be false to a certainty so they’re fair game for SF. The fun in using them is in making them *seem* plausible so the reader will suspend disbelief.

                But out there in the real world suspending disbelief rarely ends well.

                • Dunno ’bout yours, but mine do. Both have distinct personalities and interests and how tbey deal with their staff.

                  Salem Saberhagen pries open doors that aren’t fully closed and sliding windows, too. Princess Balkis knows its bedtime when the TV is turned off (she likes procedurals) and expects a last feeding before retiring. She knows how to open doors with lever locks but finds it beneath her. And she’s been known to tell on her brother when he’s been naughty.

                  They can be playful when they feel like it, friendly when they feel like and distant otherwise. They have ways to show affection and communicate disatisfaction. And especially asking for treats.

                  So yes, they both have agency and can solve problems without prompting. Figure the intelligence of a 2 year old human but alien.

                  Dogs, too.
                  Lots of dog stories, too.

  6. At last the idea is filtering out that LLMs are useful for more than chatbot queries.
    Good to hear from Ms Hogarth again.
    (Still a fan of the three jaguars.)

    https://www.amazon.com/Three-Jaguars-Comic-About-Business-ebook/dp/B0BSG51YXR/ref=tmm_kin_swatch_0?_encoding=UTF8&qid=&sr=

    Just as with the indie revolution 15 years ago, the early explorers will reap good returns from their efforts.

    Consider this:

    https://www.windowscentral.com/software-apps/ubisoft-and-nvidia-working-together-ai-generated-npcs-for-upcoming-games

    Ubisoft is one of the bigger video game developer/publishers and far from the only one looking to embed/link LLM/SLM bots in their games to provide non-player characters (NPCs) with interactive dialogue and personalities. Even outsider modders have already demonstrated this with SKYRIM.

    As pointed out in the OP, authors can feed their work to be processed by a LLM tool for a variety of creative and *non-creative* analyses. LLM biggest value doesn’t lie in derivative pastiche or fanfic level output, but in assisting creatives in improving their own output. As with word processors and spell checkers, using them will be optional but they *will* be used and used in creative ways. 😉

    PS: It has recently come to light that chatbots give better answers when told to reply as somebody else. As in “Answer as Sherlock Holmes…” or “What would Captain James T. Kirk say about…”

    In other words, the LLM model underneath is better than the chatbot interface persona knows how to use it. Interesting things are coming.

    • Consider these five alternate ways to get the LLM to bypass the chatbot “persona”:

      https://www.descript.com/blog/article/5-advanced-prompts-to-get-better-answers-from-chatgpt

      The Persona Pattern

      From now on, act as [persona]. Pay close attention to [details to focus on]. Provide outputs that [persona] would regarding the input.

      The Flipped Interaction Pattern

      From now on, I would like you to ask me questions to [do a specific task]. When you have enough information to [do the task], create [output you want].

      The Question Refinement Pattern

      From now on, when I ask a question, suggest a better version of the question to use that incorporates information specific to [use case] and ask me if I would like to use your question instead.

      The Cognitive Verifier Pattern

      When I ask you a question, generate three additional questions that would help you give a more accurate answer. When I have answered the three questions, combine the answers to produce the final answers to my original question.

      The Reflection Pattern

      When you provide an answer, please explain the reasoning and assumptions behind your response. If possible, use specific examples or evidence to support your answer of why [prompt topic] is the best. Moreover, please address any potential ambiguities or limitations in your answer, in order to provide a more complete and accurate response.

      Also, the linked VANDERBILT UNIVERSITY pdf is useful.

      We’re just scratching the surface of what the tools can do.

    • I suspect that knowing when to ignore LLM outputs will be critical in the future. (Just like knowing when to ignoring writing rules, expensive consultants, etc).

      • Of course, that requires critical thinking skills that aren’t being taught and are actively suppressed in some circles.

Comments are closed.