Is it time to hit the pause button on AI?

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.

22 thoughts on “Is it time to hit the pause button on AI?”

  1. William Buckley said he’d rather be governed by the first 2,000 names in the NY City telephone book than the faculty of Harvard.

    Likewise, I think I’d prefer AI to the very deep thinkers who want to save us from it.

    • And the thing about Pandora is that she was an AI created by spiteful gods. Adrienne Mayor, who wrote the book at that link, makes an an excellent case for the ancients thinking of her as a type of android, especially given how she’s portrayed on ancient vases and whatnot. So Pandora is twice an apt reference.

      Side note, I also recommend that book for science fantasy writers who want “magitek” inspiration. It’s eye-opening to see technological concepts we think of as modern already being imagined by people who lived thousands of years before their inventions were possible. The automaton owl, Bubo, in the Harryhausen version of “Clash of the Titans” was right in line with the androids Hephaestus made for himself. Except Hephaestus employed golden fembots and not owls 🙂 The ancients had their own versions of Iron Man and C-3P0, and their own primitive “Weta Workshop” special effects techniques.

      • Mayer is right: the ancient greeks not only understood the concept of automatons, they *made* some:

        “Few examples of automatons made prior to the 16th century remain, but numerous documents record their onetime existence. Among the earliest references is to a wooden model of a pigeon constructed by Archytas of Tarentum (flourished 400–350 BCE), a Greek friend of Plato. The bird was apparently suspended from the end of a pivoted bar, and the whole apparatus revolved by means of a jet of steam or compressed air. More complete information about other devices is found in the writings of Heron of Alexandria (flourished 1st century CE), who described devices actuated by water, falling weights, and steam.

        “Accounts of automatons in China date from as early as the 3rd century BCE, during the Han dynasty, when a mechanical orchestra was made for the emperor. By the Sui dynasty, in the 6th and 7th centuries CE, automatons had become widespread, and a book titled Shuishi tujing (“Book of Hydraulic Elegancies”) was published. In the Tang period, from the 7th to the 10th century CE, automatons continued to be popular in imperial circles. There are records of flying birds, an otter that caught fish, and figures engaged in numerous activities ranging from a monk begging to girls singing. After the Yuan period (1279–1368), the creation of automatons seems to have waned.”

        A lot is made about the greco-roman remants “preserved” by Islam but very litle about the art, literature, and engineering they destroyed because it didn’t *their* culture.

        The greek tech base was a lot more advanced than anybody before the Renaissance. If Alexander had lived things could have gotten very interesting. The only thing they lacked was steam engines and *Hero* of Alexandria was close.
        The Antikythera mechanism alone is proof of their mechanical engineering skills. Such fine craftmanship coupled with steam pwrr woukd have changed the world.

        A perfect setup for a steampunk althistory fantasy.

        Oh, and thanks for the link.

  2. The line between “what actually occurred” and The Spin, or “what whomever wants you to believe occurred,” has been badly blurred in this nation for a very long time. The general public goes along because they’re too timid and frightened to do anything else.

    America at present is a naive pup with poor vision, barking and racing headlong to attack whatever’s coming—until it gets close enough to recognize the threat is an angry bear. Whereupon it will slide to a stop, roll onto its back, and offer its belly, certain the bear will recognize it as a friend.

    Sad as it is to say, and although it will void the sacrifices of so many, the pup will have deserved exactly what it gets.

    • That certainly is the image it presents to the world.
      However, it is far from the whole story; hence the culture wars.

      Internally, the country can afford the catfights with limited damage to the economy and standard of living.

      Externally, though, it has already fostered one major war with a second likely.
      The current war is being fought by proxy cheaply with some success despite its misguided ’60’s incrementalist approach. It didn’t work in Vietnam and it is producing a bloodbath in Ukraine.

      The danger is that China is under even more stress (demographic and economic) than dying Russia and is increasingly in need of the classic “short victorious war” and may take the wrong lessons (strike now before the americans’ new systems come online) without understanding that in war they won’t be dealing with the woke, who despise the military, but with the other tribes. And that the publicly known gear is but the tip of an iceberg.

      The country is nowhere near as weak as it looks like from the outside. Note how Russia is faring against second and third tier gear. The gamble the chinese face is about the gerontocracy’s willingness to fight back and they may not understand that in times of crisis institutional culture overrides ideology.

      Best case would be for the uncrowned emperor to tend to his own house but he may not be smart enough to avoid the bloodbath. He has shown a tendency to drink his koolaid.

      • Going to have to disagree about Ukraine–it’s a lot more like Afghanistan in the 1980s than Vietnam, and with less chance of it backfiring on the US to boot.

        Hopefully the Chinese will look at Ukraine, remember the fact that an amphibious assault is more difficult than a land one by an order of magnitude, and not do something that will speed up their decline. OTOH, if they decide that the problem was Moscow doing it wrong rather than doing it at all…interesting times ahead.

        • Less of it backfiring?
          Hold your breath.
          If Ukraine does push the russians out there will be ethnic cleansing and score settling. Ukraine is as much a soviet succesor state as Russia.

          And as much as Putin postures about nuclear missiles, Ukraine has a good enough aerospace business to build drones and midrange missiles and abundant nuclear waste.

          The situation has bigtime blowback potential.

          And even in the “best case” scenario, if Russia withdraws with its tail between its legs their federation has a high probability of imploding. The oligarchs and regional bosses are already assembling private militias along the line of the WAGNER GROUP and the Chechens.

          And China, if they don’t move on Taiwan, Vladivostock and points north are more easily reached, more useful (oil!) and equal nationalist value. A russian civil war is going to lead to a lot of 19th century adventurism. (And don’t discount Turkey and Japan. Sakhalin and southern Armenia are both juicy.)

          And if Russia doesn’t implode the crackpot nationalists will succeed Putin, not Navalny.

          If the US was to intervene it needed to go all out in arming Ukraine, not just dip-dripping to bleed Russia.

          Blowback potential is high on this gamble.

          • The region between the Vistula and the Volga has been a cesspit of tribal, ethnic, religious, and other bigotry for two millenia (before then, it wasn’t densely populated enough for extensive contacts). I don’t expect that to change any time soon.

            Which, frankly, doesn’t make it all that different from the region to the west of it. Or the east of it. And definitely no different going south.

            • Emerging archaelogical data suggests humans have been doing ethnic cleansing in the entirety of northern eurasia since before the last ice age. At least as far back as 50,000 years.

              I shake my head everytime some geneticist talks of coexistence and interbreeding between “modern” humans and other homo subspecies. Yeah, right. Because peaceful coexistence is what humans are known for, right?

              (More likely: neanderthal women captured as slaves and forced bedmates. Folowed by exposure for any infant too closely resembling the mother.)

              Humans aren’t the best of neighbors.

        • A while back I read something about Taiwan’s plan to destroy the Three Gorges Dam in the event of an invasion. Pretty drastic, but who knows?

          • Not drastic, actually.
            That needs doing for the sake of China. Half their province wide floods are due to the thing. And ever year it gets closer to collapsing in the middle of rainy season. Tofu construction isn’t durable.

            Drastic is, Taiwan had a nuclear bomb program in the ’70’s and “stopped” it because the US asked nicely. (Right.)
            Word is they can assemble a nuke in a weekend.
            And Beijing is a couple minutes away by missile.

            Ukraine had 8 years to prepare for Rusia’s invasion; Taiwan has had 75+. For a while they had some pacifists willing to buy China’s promises of autonomy. Then China clamped down on Hong Kong. Bad move: they needed to take over Taiwan first, *then* crush Hong Kong.

            • Just as in Ukraine, don’t fall into the trap of presuming that there are no circular firing squads in “the 23d Province of China.” One particularly obvious one is the still burbling resentment of island natives for the highhanded Kuomintang takeover.

              Here in the West, we have a disturbing tendency to look at what we don’t understand and assume levels of unity that are all-too-readily refuted by a deeper-than-a-daily-newspaper-column look. Don’t make that mistake with Taiwan. Or Hong Kong. Or anything else in or related to “China.” We don’t do it with Texas (convenient example only, far from unique — I live in a state whose boundaries are so egregiously misdrawn that they’d make the nineteenth-century British Foreign office envious), so why should be do it with others?

              • Recent elections in Taiwan highlight that just fine.
                All those weapons the government is buying will be used. Against somebody.

                Likewise there is a lot of cheering over Japan’s overt arming (as opposed to their previous covert arming) to “face down China”. If China does collapse by 2030 (30% chance) Japan isn’t going to trash all those shiny new toys. They’ll find uses by 2050.

                Today’s “ally” is tomorrow’s problem.
                C.F., Türkiye.

  3. Like I said before, we already faced this problem. When the internet first became a thing, when Wikipedia first became popular, we already had handwringing about nefarious or wrong sources tricking the undiscerning. I welcome the growing pains of the AI, because I hope it will re-awaken people to be more discerning.

    That said, there is also the corrosive effect of not being able to trust sources. Unfortunately, our institutions and “Idiot Politicians” have worked overtime doing the lion’s share of work in undermining trust. AI as a tool for such people is already blunted as a weapon, because people today will not trust so readily in anything connected to AI as they might have before. Of course, I may be foolishly optimistic here: there are consequences for educational institutions failing so badly at teaching their students how to evaluate what they see, hear, or read.

    • Any new tech (especially if it disrupts an entrench power like Google or Sony) is going to be fiercely FUD’ed. The smart play is to ignore the end-of-the-world hzndwringers anc let the tech sink or swim on its own. (ebooks, swim–metaverse, sink).
      Less stress.

      Most of these pieces won’t age well because MS is constantly updating the product–fast. Almost daily.

      Of particular note so far:

      “Chat Tones: We’ve introduced the ability to toggle the tone of chat from “Precise”, which focuses on shorter, more search-focused answers, to “Creative” which gives responses that are longer and more descriptive. The middle setting (“Balanced”) is somewhere in-between. You’ll notice a color change in the UX between purple, blue, and green depending on which is selected. Our goal is to let you decide the type of chat behavior that best meets your needs. We’ll continue to tune this experience based on feedback.”


      “Chat Behavior: We’ve improved some chat behaviors that previously would have unnecessarily constrained responses or made them appear defensive or adversarial. Bing responses should be more engaging and provide more elaborate observations.”

      The product isn’t in general availability and at MS “beta” means “beta”, not shovelware. By the time it hits full availability it will barely resemble the product the FUD’ers decry. It’s happened before, most recently with Cloud Gaming.

    • It’s far from just the “Idiot Politicians” — the fine tradition of snake-oil salesmanship has never been restricted to politics. Neither has the unfair takedown of a competitor/competitor’s product.

      The Blues Brothers expressed it succinctly: “I didn’t lie to you. I was just… bullsh*tting you.” (And remember, it’s two ex-cons talking to each other.)

Comments are closed.