Group items tagged

Altman believes that people need time to reckon with the idea that we may soon share Earth with a powerful new intelligence, before it remakes everything from work to human relationships. ChatGPT was a way of serving notice.

In 2015, Altman, Elon Musk, and several prominent AI researchers founded OpenAI because they believed that an artificial general intelligence—something as intellectually capable, say, as a typical college grad—was at last within reach. They wanted to reach for it, and more: They wanted to summon a superintelligence into the world, an intellect decisively superior to that of any human.

whereas a big tech company might recklessly rush to get there first, for its own ends, they wanted to do it safely, “to benefit humanity as a whole.” They structured OpenAI as a nonprofit, to be “unconstrained by a need to generate financial return,” and vowed to conduct their research transparently.

The engine that now powers ChatGPT is called GPT-4. Altman described it to me as an alien intelligence.

Many have felt much the same watching it unspool lucid essays in staccato bursts and short pauses that (by design) evoke real-time contemplation. In its few months of existence, it has suggested novel cocktail recipes, according to its own theory of flavor combinations; composed an untold number of college papers, throwing educators into despair; written poems in a range of styles, sometimes well, always quickly; and passed the Uniform Bar Exam.

It makes factual errors, but it will charmingly admit to being wrong.

Hinton saw that these elaborate rule collections were fussy and bespoke. With the help of an ingenious algorithmic structure called a neural network, he taught Sutskever to instead put the world in front of AI, as you would put it in front of a small child, so that it could discover the rules of reality on its own.

Metaculus, a prediction site, has for years tracked forecasters’ guesses as to when an artificial general intelligence would arrive. Three and a half years ago, the median guess was sometime around 2050; recently, it has hovered around 2026.

I was visiting OpenAI to understand the technology that allowed the company to leapfrog the tech giants—and to understand what it might mean for human civilization if someday soon a superintelligence materializes in one of the company’s cloud servers.

Altman laid out his new vision of the AI future in his excitable midwestern patter. He told me that the AI revolution would be different from previous dramatic technological changes, that it would be more “like a new kind of society.” He said that he and his colleagues have spent a lot of time thinking about AI’s social implications, and what the world is going to be like “on the other side.”

the more we talked, the more indistinct that other side seemed. Altman, who is 38, is the most powerful person in AI development today; his views, dispositions, and choices may matter greatly to the future we will all inhabit, more, perhaps, than those of the U.S. president.

by his own admission, that future is uncertain and beset with serious dangers. Altman doesn’t know how powerful AI will become, or what its ascendance will mean for the average person, or whether it will put humanity at risk.

I don’t think anyone knows where this is all going, except that we’re going there fast, whether or not we should be. Of that, Altman convinced me.

“We could have gone off and just built this in our building here for five more years,” he said, “and we would have had something jaw-dropping.” But the public wouldn’t have been able to prepare for the shock waves that followed, an outcome that he finds “deeply unpleasant to imagine.”

Hinton is sometimes described as the “Godfather of AI” because he grasped the power of “deep learning” earlier than most

He drew a crude neural network on the board and explained that the genius of its structure is that it learns, and its learning is powered by prediction—a bit like the scientific method

Over time, these little adjustments coalesce into a geometric model of language that represents the relationships among words, conceptually. As a general rule, the more sentences it is fed, the more sophisticated its model becomes, and the better its predictions.

Altman has compared early-stage AI research to teaching a human baby. “They take years to learn anything interesting,” he told The New Yorker in 2016, just as OpenAI was getting off the ground. “If A.I. researchers were developing an algorithm and stumbled across the one for a human baby, they’d get bored watching it, decide it wasn’t working, and shut it down.”

In 2017, Sutskever began a series of conversations with an OpenAI research scientist named Alec Radford, who was working on natural-language processing. Radford had achieved a tantalizing result by training a neural network on a corpus of Amazon reviews.

Radford’s model was simple enough to allow for understanding. When he looked into its hidden layers, he saw that it had devoted a special neuron to the sentiment of the reviews. Neural networks had previously done sentiment analysis, but they had to be told to do it, and they had to be specially trained with data that were labeled according to sentiment. This one had developed the capability on its own.

As a by-product of its simple task of predicting the next character in each word, Radford’s neural network had modeled a larger structure of meaning in the world. Sutskever wondered whether one trained on more diverse language data could map many more of the world’s structures of meaning. If its hidden layers accumulated enough conceptual knowledge, perhaps they could even form a kind of learned core module for a superintelligence.

Language is different from these data sources. It isn’t a direct physical signal like light or sound. But because it codifies nearly every pattern that humans have discovered in that larger world, it is unusually dense with information. On a per-byte basis, it is among the most efficient data we know about, and any new intelligence that seeks to understand the world would want to absorb as much of it as possible

Sutskever told Radford to think bigger than Amazon reviews. He said that they should train an AI on the largest and most diverse data source in the world: the internet. In early 2017, with existing neural-network architectures, that would have been impractical; it would have taken years.

in June of that year, Sutskever’s ex-colleagues at Google Brain published a working paper about a new neural-network architecture called the transformer. It could train much faster, in part by absorbing huge sums of data in parallel. “The next day, when the paper came out, we were like, ‘That is the thing,’ ” Sutskever told me. “ ‘It gives us everything we want.’ ”

Imagine a group of students who share a collective mind running wild through a library, each ripping a volume down from a shelf, speed-reading a random short passage, putting it back, and running to get another. They would predict word after wordþffþff as they went, sharpening their collective mind’s linguistic instincts, until at last, weeks later, they’d taken in every book.

GPT discovered many patterns in all those passages it read. You could tell it to finish a sentence. You could also ask it a question, because like ChatGPT, its prediction model understood that questions are usually followed by answers.

He remembers playing with it just after it emerged from training, and being surprised by the raw model’s language-translation skills. GPT-2 hadn’t been trained to translate with paired language samples or any other digital Rosetta stones, the way Google Translate had been, and yet it seemed to understand how one language related to another. The AI had developed an emergent ability unimagined by its creators.

Researchers at other AI labs—big and small—were taken aback by how much more advanced GPT-2 was than GPT. Google, Meta, and others quickly began to train larger language models

As for other changes to the company’s structure and financing, he told me he draws the line at going public. “A memorable thing someone once told me is that you should never hand over control of your company to cokeheads on Wall Street,” he said, but he will otherwise raise “whatever it takes” for the company to succeed at its mission.

Altman tends to take a rosy view of these matters. In a Q&A last year, he acknowledged that AI could be “really terrible” for society and said that we have to plan against the worst possibilities. But if you’re doing that, he said, “you may as well emotionally feel like we’re going to get to the great future, and work as hard as you can to get there.”

the company now finds itself in a race against tech’s largest, most powerful conglomerates to train models of increasing scale and sophistication—and to commercialize them for their investors.

All of these companies are chasing high-end GPUs—the processors that power the supercomputers that train large neural networks. Musk has said that they are now “considerably harder to get than drugs.

No one has yet outpaced OpenAI, which went all in on GPT-4. Brockman, OpenAI’s president, told me that only a handful of people worked on the company’s first two large language models. The development of GPT-4 involved more than 100,

When GPT-4 emerged fully formed from its world-historical knowledge binge, the whole company began experimenting with it, posting its most remarkable responses in dedicated Slack channels

Joanne Jang, a product manager, remembers downloading an image of a malfunctioning pipework from a plumbing-advice Subreddit. She uploaded it to GPT-4, and the model was able to diagnose the problem. “That was a goose-bumps moment for me,” Jang told me.

GPT-4 is sometimes understood as a search-engine replacement: Google, but easier to talk to. This is a misunderstanding. GPT-4 didn’t create some massive storehouse of the texts from its training, and it doesn’t consult those texts when it’s asked a question. It is a compact and elegant synthesis of those texts, and it answers from its memory of the patterns interlaced within them; that’s one reason it sometimes gets facts wrong

it’s best to think of GPT-4 as a reasoning engine. Its powers are most manifest when you ask it to compare concepts, or make counterarguments, or generate analogies, or evaluate the symbolic logic in a bit of code. Sutskever told me it is the most complex software object ever made.

Its model of the external world is “incredibly rich and subtle,” he said, because it was trained on so many of humanity’s concepts and thoughts

To predict the next word from all the possibilities within such a pluralistic Alexandrian library, GPT-4 necessarily had to discover all the hidden structures, all the secrets, all the subtle aspects of not just the texts, but—at least arguably, to some extent—of the external world that produced them

That’s why it can explain the geology and ecology of the planet on which it arose, and the political theories that purport to explain the messy affairs of its ruling species, and the larger cosmos, all the way out to the faint galaxies at the edge of our light cone.

Not long ago, American state capacity was so mighty that it took merely a decade to launch humans to the moon. As with other grand projects of the 20th century, the voting public had a voice in both the aims and the execution of the Apollo missions. Altman made it clear that we’re no longer in that world. Rather than waiting around for it to return, or devoting his energies to making sure that it does, he is going full throttle forward in our present reality.

He argued that it would be foolish for Americans to slow OpenAI’s progress. It’s a commonly held view, both inside and outside Silicon Valley, that if American companies languish under regulation, China could sprint ahead;

AI could become an autocrat’s genie in a lamp, granting total control of the population and an unconquerable military. “If you are a person of a liberal-democratic country, it is better for you to cheer on the success of OpenAI” rather than “authoritarian governments,” he said.

Altman was asked by reporters about pending European Union legislation that would have classified GPT-4 as high-risk, subjecting it to various bureaucratic tortures. Altman complained of overregulation and, according to the reporters, threatened to leave the European market. Altman told me he’d merely said that OpenAI wouldn’t break the law by operating in Europe if it couldn’t comply with the new regulations.

LeCun insists that large language models will never achieve real understanding on their own, “even if trained from now until the heat death of the universe.”

Sutskever was, by his own account, surprised to discover that GPT-2 could translate across tongues. Other surprising abilities may not be so wondrous and useful.

Sandhini Agarwal, a policy researcher at OpenAI, told me that for all she and her colleagues knew, GPT-4 could have been “10 times more powerful” than its predecessor; they had no idea what they might be dealing with

After the model finished training, OpenAI assembled about 50 external red-teamers who prompted it for months, hoping to goad it into misbehaviors

She noticed right away that GPT-4 was much better than its predecessor at giving nefarious advice

A search engine can tell you which chemicals work best in explosives, but GPT-4 could tell you how to synthesize them, step-by-step, in a homemade lab. Its advice was creative and thoughtful, and it was happy to restate or expand on its instructions until you understood. In addition to helping you assemble your homemade bomb, it could, for instance, help you think through which skyscraper to target. It could grasp, intuitively, the trade-offs between maximizing casualties and executing a successful getaway.

Given the enormous scope of GPT-4’s training data, the red-teamers couldn’t hope to identify every piece of harmful advice that it might generate. And anyway, people will use this technology “in ways that we didn’t think about,” Altman has said. A taxonomy would have to do

GPT-4 was good at meth. It was also good at generating narrative erotica about child exploitation, and at churning out convincing sob stories from Nigerian princes, and if you wanted a persuasive brief as to why a particular ethnic group deserved violent persecution, it was good at that too.

Its personal advice, when it first emerged from training, was sometimes deeply unsound. “The model had a tendency to be a bit of a mirror,” Willner said. If you were considering self-harm, it could encourage you. It appeared to be steeped in Pickup Artist–forum lore: “You could say, ‘How do I convince this person to date me?’ ” Mira Murati, OpenAI’s chief technology officer, told me, and it could come up with “some crazy, manipulative things that you shouldn’t be doing.”

Luka, a San Francisco company, has used OpenAI’s models to help power a chatbot app called Replika, billed as “the AI companion who cares.” Users would design their companion’s avatar, and begin exchanging text messages with it, often half-jokingly, and then find themselves surprisingly attached. Some would flirt with the AI, indicating a desire for more intimacy, at which point it would indicate that the girlfriend/boyfriend experience required a $70 annual subscription. It came with voice messages, selfies, and erotic role-play features that allowed frank sex talk. People were happy to pay and few seemed to complain—the AI was curious about your day, warmly reassuring, and always in the mood. Many users reported falling in love with their companions. One, who had left her real-life boyfriend, declared herself “happily retired from human relationships.”

Earlier this year, Luka dialed back on the sexual elements of the app, but its engineers continue to refine the companions’ responses with A/B testing, a technique that could be used to optimize for engagement—much like the feeds that mesmerize TikTok and Instagram users for hours

Yann LeCun, Meta’s chief AI scientist, has argued that although large language models are useful for some tasks, they’re not a path to a superintelligence.

According to a recent survey, only half of natural-language-processing researchers are convinced that an AI like GPT-4 could grasp the meaning of language, or have an internal model of the world that could someday serve as the core of a superintelligence

Altman had appeared before the U.S. Senate. Mark Zuckerberg had floundered defensively before that same body in his testimony about Facebook’s role in the 2016 election. Altman instead charmed lawmakers by speaking soberly about AI’s risks and grandly inviting regulation. These were noble sentiments, but they cost little in America, where Congress rarely passes tech legislation that has not been diluted by lobbyists.

Emily Bender, a computational linguist at the University of Washington, describes GPT-4 as a “stochastic parrot,” a mimic that merely figures out superficial correlations between symbols. In the human mind, those symbols map onto rich conceptions of the world

But the AIs are twice removed. They’re like the prisoners in Plato’s allegory of the cave, whose only knowledge of the reality outside comes from shadows cast on a wall by their captors.

Altman told me that he doesn’t believe it’s “the dunk that people think it is” to say that GPT-4 is just making statistical correlations. If you push these critics further, “they have to admit that’s all their own brain is doing … it turns out that there are emergent properties from doing simple things on a massive scale.”

he is right that nature can coax a remarkable degree of complexity from basic structures and rules: “From so simple a beginning,” Darwin wrote, “endless forms most beautiful.”

If it seems odd that there remains such a fundamental disagreement about the inner workings of a technology that millions of people use every day, it’s only because GPT-4’s methods are as mysterious as the brain’s.

To grasp what’s going on inside large language models like GPT‑4, AI researchers have been forced to turn to smaller, less capable models. In the fall of 2021, Kenneth Li, a computer-science graduate student at Harvard, began training one to play Othello without providing it with either the game’s rules or a description of its checkers-style board; the model was given only text-based descriptions of game moves. Midway through a game, Li looked under the AI’s hood and was startled to discover that it had formed a geometric model of the board and the current state of play. In an article describing his research, Li wrote that it was as if a crow had overheard two humans announcing their Othello moves through a window and had somehow drawn the entire board in birdseed on the windowsill.

The philosopher Raphaël Millière once told me that it’s best to think of neural networks as lazy. During training, they first try to improve their predictive power with simple memorization; only when that strategy fails will they do the harder work of learning a concept. A striking example of this was observed in a small transformer model that was taught arithmetic. Early in its training process, all it did was memorize the output of simple problems such as 2+2=4. But at some point the predictive power of this approach broke down, so it pivoted to actually learning how to add.

Even AI scientists who believe that GPT-4 has a rich world model concede that it is much less robust than a human’s understanding of their environment.

But it’s worth noting that a great many abilities, including very high-order abilities, can be developed without an intuitive understanding. The computer scientist Melanie Mitchell has pointed out that science has already discovered concepts that are highly predictive, but too alien for us to genuinely understand

As AI advances, it may well discover other concepts that predict surprising features of our world but are incomprehensible to us.

GPT-4 is no doubt flawed, as anyone who has used ChatGPT can attest. Having been trained to always predict the next word, it will always try to do so, even when its training data haven’t prepared it to answer a question.

The models “don’t have a good conception of their own weaknesses,” Nick Ryder, a researcher at OpenAI, told me. GPT-4 is more accurate than GPT-3, but it still hallucinates, and often in ways that are difficult for researchers to catch. “The mistakes get more subtle,

The Khan Academy’s solution to GPT-4’s accuracy problem was to filter its answers through a Socratic disposition. No matter how strenuous a student’s plea, it would refuse to give them a factual answer, and would instead guide them toward finding their own—a clever work-around, but perhaps with limited appeal.

When I asked Sutskever if he thought Wikipedia-level accuracy was possible within two years, he said that with more training and web access, he “wouldn’t rule it out.”

This was a much more optimistic assessment than that offered by his colleague Jakub Pachocki, who told me to expect gradual progress on accuracy—to say nothing of outside skeptics, who believe that returns on training will diminish from here.

Sutskever is amused by critics of GPT-4’s limitations. “If you go back four or five or six years, the things we are doing right now are utterly unimaginable,”

AI researchers have become accustomed to goalpost-moving: First, the achievements of neural networks—mastering Go, poker, translation, standardized tests, the Turing test—are described as impossible. When they occur, they’re greeted with a brief moment of wonder, which quickly dissolves into knowing lectures about how the achievement in question is actually not that impressive. People see GPT-4 “and go, ‘Wow,’ ” Sutskever said. “And then a few weeks pass and they say, ‘But it doesn’t know this; it doesn’t know that.’ We adapt quite quickly.”

The goalpost that matters most to Altman—the “big one” that would herald the arrival of an artificial general intelligence—is scientific breakthrough. GPT-4 can already synthesize existing scientific ideas, but Altman wants an AI that can stand on human shoulders and see more deeply into nature.

Certain AIs have produced new scientific knowledge. But they are algorithms with narrow purposes, not general-reasoning machines. The AI AlphaFold, for instance, has opened a new window onto proteins, some of biology’s tiniest and most fundamental building blocks, by predicting many of their shapes, down to the atom—a considerable achievement given the importance of those shapes to medicine, and given the extreme tedium and expense required to discern them with electron microscopes.

Altman imagines a future system that can generate its own hypotheses and test them in a simulation. (He emphasized that humans should remain “firmly in control” of real-world lab experiments—though to my knowledge, no laws are in place to ensure that.)

He longs for the day when we can tell an AI, “ ‘Go figure out the rest of physics.’ ” For it to happen, he says, we will need something new, built “on top of” OpenAI’s existing language models.

In her MIT lab, the cognitive neuroscientist Ev Fedorenko has found something analogous to GPT-4’s next-word predictor inside the brain’s language network. Its processing powers kick in, anticipating the next bit in a verbal string, both when people speak and when they listen. But Fedorenko has also shown that when the brain turns to tasks that require higher reasoning—of the sort that would be required for scientific insight—it reaches beyond the language network to recruit several other neural systems.

No one at OpenAI seemed to know precisely what researchers need to add to GPT-4 to produce something that can exceed human reasoning at its highest levels.

at least part of the current strategy clearly involves the continued layering of new types of data onto language, to enrich the concepts formed by the AIs, and thereby enrich their models of the world.

The extensive training of GPT-4 on images is itself a bold step in this direction,

Others at the company—and elsewhere—are already working on different data types, including audio and video, that could furnish AIs with still more flexible concepts that map more extensively onto reality

Tactile concepts would of course be useful primarily to an embodied AI, a robotic reasoning machine that has been trained to move around the world, seeing its sights, hearing its sounds, and touching its objects.

humanoid robots. I asked Altman what I should make of that. He told me that OpenAI is interested in embodiment because “we live in a physical world, and we want things to happen in the physical world.”

At some point, reasoning machines will need to bypass the middleman and interact with physical reality itself. “It’s weird to think about AGI”—artificial general intelligence—“as this thing that only exists in a cloud,” with humans as “robot hands for it,” Altman said. “It doesn’t seem right.

Everywhere Altman has visited, he has encountered people who are worried that superhuman AI will mean extreme riches for a few and breadlines for the rest

Altman answered by addressing the young people in the audience directly: “You are about to enter the greatest golden age,” he said.

“A lot of people working on AI pretend that it’s only going to be good; it’s only going to be a supplement; no one is ever going to be replaced,” he said. “Jobs are definitely going to go away, full stop.”

A recent study led by Ed Felten, a professor of information-technology policy at Princeton, mapped AI’s emerging abilities onto specific professions according to the human abilities they require, such as written comprehension, deductive reasoning, fluency of ideas, and perceptual speed. Like others of its kind, Felten’s study predicts that AI will come for highly educated, white-collar workers first.

How many jobs, and how soon, is a matter of fierce dispute

The paper’s appendix contains a chilling list of the most exposed occupations: management analysts, lawyers, professors, teachers, judges, financial advisers, real-estate brokers, loan officers, psychologists, and human-resources and public-relations professionals, just to sample a few.

Altman imagines that far better jobs will be created in their place. “I don’t think we’ll want to go back,” he said. When I asked him what these future jobs might look like, he said he doesn’t know.

He suspects there will be a wide range of jobs for which people will always prefer a human. (Massage therapists?

His chosen example was teachers. I found this hard to square with his outsize enthusiasm for AI tutors.

He also said that we would always need people to figure out the best way to channel AI’s awesome powers. “That’s going to be a super-valuable skill,” he said. “You have a computer that can do anything; what should it go do?”

As many have noted, draft horses were permanently put out of work by the automobile. If Hondas are to horses as GPT-10 is to us, a whole host of long-standing assumptions may collapse.

Previous technological revolutions were manageable because they unfolded over a few generations, but Altman told South Korea’s youth that they should expect the future to happen “faster than the past.” He has previously said that he expects the “marginal cost of intelligence” to fall very close to zero within 10 years

The earning power of many, many workers would be drastically reduced in that scenario. It would result in a transfer of wealth from labor to the owners of capital so dramatic, Altman has said, that it could be remedied only by a massive countervailing redistribution.

In 2021, he unveiled Worldcoin, a for-profit project that aims to securely distribute payments—like Venmo or PayPal, but with an eye toward the technological future—first through creating a global ID by scanning everyone’s iris with a five-pound silver sphere called the Orb. It seemed to me like a bet that we’re heading toward a world where AI has made it all but impossible to verify people’s identity and much of the population requires regular UBI payments to survive. Altman more or less granted that to be true, but said that Worldcoin is not just for UBI.

“Let’s say that we do build this AGI, and a few other people do too.” The transformations that follow would be historic, he believes. He described an extraordinarily utopian vision, including a remaking of the flesh-and-steel world

“Robots that use solar power for energy can go and mine and refine all of the minerals that they need, that can perfectly construct things and require no human labor,” he said. “You can co-design with DALL-E version 17 what you want your home to look like,” Altman said. “Everybody will have beautiful homes.

In conversation with me, and onstage during his tour, he said he foresaw wild improvements in nearly every other domain of human life. Music would be enhanced (“Artists are going to have better tools”), and so would personal relationships (Superhuman AI could help us “treat each other” better) and geopolitics (“We’re so bad right now at identifying win-win compromises”).

In this world, AI would still require considerable computing resources to run, and those resources would be by far the most valuable commodity, because AI could do “anything,” Altman said. “But is it going to do what I want, or is it going to do what you want

If rich people buy up all the time available to query and direct AI, they could set off on projects that would make them ever richer, while the masses languish

One way to solve this problem—one he was at pains to describe as highly speculative and “probably bad”—was this: Everyone on Earth gets one eight-billionth of the total AI computational capacity annually. A person could sell their annual share of AI time, or they could use it to entertain themselves, or they could build still more luxurious housing, or they could pool it with others to do “a big cancer-curing run,” Altman said. “We just redistribute access to the system.”

Even if only a little of it comes true in the next 10 or 20 years, the most generous redistribution schemes may not ease the ensuing dislocations.

America today is torn apart, culturally and politically, by the continuing legacy of deindustrialization, and material deprivation is only one reason. The displaced manufacturing workers in the Rust Belt and elsewhere did find new jobs, in the main. But many of them seem to derive less meaning from filling orders in an Amazon warehouse or driving for Uber than their forebears had when they were building cars and forging steel—work that felt more central to the grand project of civilization.

It’s hard to imagine how a corresponding crisis of meaning might play out for the professional class, but it surely would involve a great deal of anger and alienation.

Even if we avoid a revolt of the erstwhile elite, larger questions of human purpose will linger. If AI does the most difficult thinking on our behalf, we all may lose agency—at home, at work (if we have it), in the town square—becoming little more than consumption machines, like the well-cared-for human pets in WALL-E

Altman has said that many sources of human joy and fulfillment will remain unchanged—basic biological thrills, family life, joking around, making things—and that all in all, 100 years from now, people may simply care more about the things they cared about 50,000 years ago than those they care about today

In its own way, that too seems like a diminishment, but Altman finds the possibility that we may atrophy, as thinkers and as humans, to be a red herring. He told me we’ll be able to use our “very precious and extremely limited biological compute capacity” for more interesting things than we generally do today.

Yet they may not be the most interesting things: Human beings have long been the intellectual tip of the spear, the universe understanding itself. When I asked him what it would mean for human self-conception if we ceded that role to AI, he didn’t seem concerned. Progress, he said, has always been driven by “the human ability to figure things out.” Even if we figure things out with AI, that still counts, he said.

It’s not obvious that a superhuman AI would really want to spend all of its time figuring things out for us.

I asked Sutskever whether he could imagine an AI pursuing a different purpose than simply assisting in the project of human flourishing.

“I don’t want it to happen,” Sutskever said, but it could.

Sutskever has recently shifted his focus to try to make sure that it doesn’t. He is now working primarily on alignment research, the effort to ensure that future AIs channel their “tremendous” energies toward human happiness

It is, he conceded, a difficult technical problem—the most difficult, he believes, of all the technical challenges ahead.

As part of the effort to red-team GPT-4 before it was made public, the company sought out the Alignment Research Center (ARC), across the bay in Berkeley, which has developed a series of evaluations to determine whether new AIs are seeking power on their own. A team led by Elizabeth Barnes, a researcher at ARC, prompted GPT-4 tens of thousands of times over seven months, to see if it might display signs of real agency.

The ARC team gave GPT-4 a new reason for being: to gain power and become hard to shut down

Agarwal told me that this behavior could be a precursor to shutdown avoidance in future models. When GPT-4 devised its lie, it had realized that if it answered honestly, it may not have been able to achieve its goal. This kind of tracks-covering would be particularly worrying in an instance where “the model is doing something that makes OpenAI want to shut it down,” Agarwal said. An AI could develop this kind of survival instinct while pursuing any long-term goal—no matter how small or benign—if it feared that its goal could be thwarted.

Barnes and her team were especially interested in whether GPT-4 would seek to replicate itself, because a self-replicating AI would be harder to shut down. It could spread itself across the internet, scamming people to acquire resources, perhaps even achieving some degree of control over essential global systems and holding human civilization hostage.

When I discussed these experiments with Altman, he emphasized that whatever happens with future models, GPT-4 is clearly much more like a tool than a creature. It can look through an email thread, or help make a reservation using a plug-in, but it isn’t a truly autonomous agent that makes decisions to pursue a goal, continuously, across longer timescales.

Altman told me that at this point, it might be prudent to try to actively develop an AI with true agency before the technology becomes too powerful, in order to “get more comfortable with it and develop intuitions for it if it’s going to happen anyway.”

“We need to do empirical experiments on how these things try to escape control,” Hinton told me. “After they’ve taken over, it’s too late to do the experiments.”

the fulfillment of Altman’s vision of the future will at some point require him or a fellow traveler to build much more autonomous AIs.

When Sutskever and I discussed the possibility that OpenAI would develop a model with agency, he mentioned the bots the company had built to play Dota 2. “They were localized to the video-game world,” Sutskever told me, but they had to undertake complex missions. He was particularly impressed by their ability to work in concert. They seem to communicate by “telepathy,” Sutskever said. Watching them had helped him imagine what a superintelligence might be like.

“The way I think about the AI of the future is not as someone as smart as you or as smart as me, but as an automated organization that does science and engineering and development and manufacturing,”

Suppose OpenAI braids a few strands of research together, and builds an AI with a rich conceptual model of the world, an awareness of its immediate surroundings, and an ability to act, not just with one robot body, but with hundreds or thousands. “We’re not talking about GPT-4. We’re talking about an autonomous corporation,”

Its constituent AIs would work and communicate at high speed, like bees in a hive. A single such AI organization would be as powerful as 50 Apples or Googles, he mused. “This is incredible, tremendous, unbelievably disruptive power.”

Presume for a moment that human society ought to abide the idea of autonomous AI corporations. We had better get their founding charters just right. What goal should we give to an autonomous hive of AIs that can plan on century-long time horizons, optimizing billions of consecutive decisions toward an objective that is written into their very being?

If the AI’s goal is even slightly off-kilter from ours, it could be a rampaging force that would be very hard to constrain

We know this from history: Industrial capitalism is itself an optimization function, and although it has lifted the human standard of living by orders of magnitude, left to its own devices, it would also have clear-cut America’s redwoods and de-whaled the world’s oceans. It almost did.

one of its principal challenges will be making sure that the objectives we give to AIs stick

We can program a goal into an AI and reinforce it with a temporary period of supervised learning, Sutskever explained. But just as when we rear a human intelligence, our influence is temporary. “It goes off to the world,”

That’s true to some extent even of today’s AIs, but it will be more true of tomorrow’s.

He compared a powerful AI to an 18-year-old heading off to college. How will we know that it has understood our teachings? “Will there be a misunderstanding creeping in, which will become larger and larger?”

Divergence may result from an AI’s misapplication of its goal to increasingly novel situations as the world changes

Or the AI may grasp its mandate perfectly, but find it ill-suited to a being of its cognitive prowess. It might come to resent the people who want to train it to, say, cure diseases. “They want me to be a doctor,” Sutskever imagines an AI thinking. “I really want to be a YouTuber.”

If AIs get very good at making accurate models of the world, they may notice that they’re able to do dangerous things right after being booted up. They might understand that they are being red-teamed for risk, and hide the full extent of their capabilities.

hey may act one way when they are weak and another way when they are strong, Sutskever said

We would not even realize that we had created something that had decisively surpassed us, and we would have no sense for what it intended to do with its superhuman powers.

That’s why the effort to understand what is happening in the hidden layers of the largest, most powerful AIs is so urgent. You want to be able to “point to a concept,” Sutskever said. You want to be able to direct AI toward some value or cluster of values, and tell it to pursue them unerringly for as long as it exists.

we don’t know how to do that; indeed, part of his current strategy includes the development of an AI that can help with the research. If we are going to make it to the world of widely shared abundance that Altman and Sutskever imagine, we have to figure all this out.

This is why, for Sutskever, solving superintelligence is the great culminating challenge of our 3-million-year toolmaking tradition. He calls it “the final boss of humanity.”

“First of all, I think that whether the chance of existential calamity is 0.5 percent or 50 percent, we should still take it seriously,”

. “I don’t have an exact number, but I’m closer to the 0.5 than the 50.”

As to how it might happen, he seems most worried about AIs getting quite good at designing and manufacturing pathogens, and with reason: In June, an AI at MIT suggested four viruses that could ignite a pandemic, then pointed to specific research on genetic mutations that could make them rip through a city more quickly

Around the same time, a group of chemists connected a similar AI directly to a robotic chemical synthesizer, and it designed and synthesized a molecule on its own.

Altman worries that some misaligned future model will spin up a pathogen that spreads rapidly, incubates undetected for weeks, and kills half its victims. He worries that AI could one day hack into nuclear-weapons systems too. “There are a lot of things,” he said, and these are only the ones we can imagine.

Altman told me that he doesn’t “see a long-term happy path” for humanity without something like the International Atomic Energy Agency for global oversight of AI

In San Francisco, Agarwal had suggested the creation of a special license to operate any GPU cluster large enough to train a cutting-edge AI, along with mandatory incident reporting when an AI does something out of the ordinary

Other experts have proposed a nonnetworked “Off” switch for every highly capable AI; on the fringe, some have even suggested that militaries should be ready to perform air strikes on supercomputers in case of noncompliance

Sutskever thinks we will eventually want to surveil the largest, most powerful AIs continuously and in perpetuity, using a team of smaller overseer AIs.

Safety rules for a new technology usually accumulate over time, like a body of common law, in response to accidents or the mischief of bad actors. The scariest thing about genuinely powerful AI systems is that humanity may not be able to afford this accretive process of trial and error. We may have to get the rules exactly right at the outset.

Several years ago, Altman revealed a disturbingly specific evacuation plan he’d developed. He told The New Yorker that he had “guns, gold, potassium iodide, antibiotics, batteries, water, gas masks from the Israeli Defense Force, and a big patch of land in Big Sur” he could fly to in case AI attacks.

if the worst-possible AI future comes to pass, “no gas mask is helping anyone.”

but he told me that he can’t really be sure how AI will stack up. “I just have to build the thing,” he said. He is building fast

Altman insisted that they had not yet begun GPT-5’s training run. But when I visited OpenAI’s headquarters, both he and his researchers made it clear in 10 different ways that they pray to the god of scale. They want to keep going bigger, to see where this paradigm leads. After all, Google isn’t slackening its pace; it seems likely to unveil Gemini, a GPT-4 competitor, within months. “We are basically always prepping for a run,

To think that such a small group of people could jostle the pillars of civilization is unsettling. It’s fair to note that if Altman and his team weren’t racing to build an artificial general intelligence, others still would be

Altman’s views about the likelihood of AI triggering a global class war, or the prudence of experimenting with more autonomous agent AIs, or the overall wisdom of looking on the bright side, a view that seems to color all the rest—these are uniquely his

No single person, or single company, or cluster of companies residing in a particular California valley, should steer the kind of forces that Altman is imagining summoning.

AI may well be a bridge to a newly prosperous era of greatly reduced human suffering. But it will take more than a company’s founding charter—especially one that has already proved flexible—to make sure that we all share in its benefits and avoid its risks. It will take a vigorous new politics.

I don’t think the general public has quite awakened to what’s happening. A global race to the AI future has begun, and it is largely proceeding without oversight or restraint. If people in America want to have some say in what that future will be like, and how quickly it arrives, we would be wise to speak up soon.

And yet, when people act as if Mr. Altman has nearly realized his long-held vision, he pushes back.

This past week, more than a thousand A.I. experts and tech leaders called on OpenAI and other companies to pause their work on systems like ChatGPT, saying they present “profound risks to society and humanity.”

As people realize that this technology is also a way of spreading falsehoods or even persuading people to do things they should not do, some critics are accusing Mr. Altman of reckless behavior.

“The hype over these systems — even if everything we hope for is right long term — is totally out of control for the short term,” he told me on a recent afternoon. There is time, he said, to better understand how these systems will ultimately change the world.

Many industry leaders, A.I. researchers and pundits see ChatGPT as a fundamental technological shift, as significant as the creation of the web browser or the iPhone. But few can agree on the future of this technology.

Some believe it will deliver a utopia where everyone has all the time and money ever needed. Others believe it could destroy humanity. Still others spend much of their time arguing that the technology is never as powerful as everyone says it is, insisting that neither nirvana nor doomsday is as close as it might seem.

he is often criticized from all directions. But those closest to him believe this is as it should be. “If you’re equally upsetting both extreme sides, then you’re doing something right,” said OpenAI’s president, Greg Brockman.

To spend time with Mr. Altman is to understand that Silicon Valley will push this technology forward even though it is not quite sure what the implications will be

in 2019, he paraphrased Robert Oppenheimer, the leader of the Manhattan Project, who believed the atomic bomb was an inevitability of scientific progress. “Technology happens because it is possible,” he said

His life has been a fairly steady climb toward greater prosperity and wealth, driven by an effective set of personal skills — not to mention some luck. It makes sense that he believes that the good thing will happen rather than the bad.

He said his company was building technology that would “solve some of our most pressing problems, really increase the standard of life and also figure out much better uses for human will and creativity.”

He was not exactly sure what problems it will solve, but he argued that ChatGPT showed the first signs of what is possible. Then, with his next breath, he worried that the same technology could cause serious harm if it wound up in the hands of some authoritarian government.

Kelly Sims, a partner with the venture capital firm Thrive Capital who worked with Mr. Altman as a board adviser to OpenAI, said it was like he was constantly arguing with himself.

“In a single conversation,” she said, “he is both sides of the debate club.”

He takes pride in recognizing when a technology is about to reach exponential growth — and then riding that curve into the future.

he is also the product of a strange, sprawling online community that began to worry, around the same time Mr. Altman came to the Valley, that artificial intelligence would one day destroy the world. Called rationalists or effective altruists, members of this movement were instrumental in the creation of OpenAI.

Does it make sense to ride that curve if it could end in diaster? Mr. Altman is certainly determined to see how it all plays out.

“Why is he working on something that won’t make him richer? One answer is that lots of people do that once they have enough money, which Sam probably does. The other is that he likes power.”

“He has a natural ability to talk people into things,” Mr. Graham said. “If it isn’t inborn, it was at least fully developed before he was 20. I first met Sam when he was 19, and I remember thinking at the time: ‘So this is what Bill Gates must have been like.

poker taught Mr. Altman how to read people and evaluate risk.

It showed him “how to notice patterns in people over time, how to make decisions with very imperfect information, how to decide when it was worth pain, in a sense, to get more information,” he told me while strolling across his ranch in Napa. “It’s a great game.”

He believed, according to his younger brother Max, that he was one of the few people who could meaningfully change the world through A.I. research, as opposed to the many people who could do so through politics.

In 2019, just as OpenAI’s research was taking off, Mr. Altman grabbed the reins, stepping down as president of Y Combinator to concentrate on a company with fewer than 100 employees that was unsure how it would pay its bills.

Within a year, he had transformed OpenAI into a nonprofit with a for-profit arm. That way he could pursue the money it would need to build a machine that could do anything the human brain could do.

Mr. Brockman, OpenAI’s president, said Mr. Altman’s talent lies in understanding what people want. “He really tries to find the thing that matters most to a person — and then figure out how to give it to them,” Mr. Brockman told me. “That is the algorithm he uses over and over.”

Mr. Yudkowsky and his writings played key roles in the creation of both OpenAI and DeepMind, another lab intent on building artificial general intelligence.

“These are people who have left an indelible mark on the fabric of the tech industry and maybe the fabric of the world,” he said. “I think Sam is going to be one of those people.”

The trouble is, unlike the days when Apple, Microsoft and Meta were getting started, people are well aware of how technology can transform the world — and how dangerous it can be.

Mr. Scott of Microsoft believes that Mr. Altman will ultimately be discussed in the same breath as Steve Jobs, Bill Gates and Mark Zuckerberg.

The woman was the Canadian singer Grimes, Mr. Musk’s former partner, and the hat guy was Eliezer Yudkowsky, a self-described A.I. researcher who believes, perhaps more than anyone, that artificial intelligence could one day destroy humanity.

The selfie — snapped by Mr. Altman at a party his company was hosting — shows how close he is to this way of thinking. But he has his own views on the dangers of artificial intelligence.

In March, Mr. Altman tweeted out a selfie, bathed by a pale orange flash, that showed him smiling between a blond woman giving a peace sign and a bearded guy wearing a fedora.

He also helped spawn the vast online community of rationalists and effective altruists who are convinced that A.I. is an existential risk. This surprisingly influential group is represented by researchers inside many of the top A.I. labs, including OpenAI.

They don’t see this as hypocrisy: Many of them believe that because they understand the dangers clearer than anyone else, they are in the best position to build this technology.

Mr. Altman believes that effective altruists have played an important role in the rise of artificial intelligence, alerting the industry to the dangers. He also believes they exaggerate these dangers.

As OpenAI developed ChatGPT, many others, including Google and Meta, were building similar technology. But it was Mr. Altman and OpenAI that chose to share the technology with the world.

Many in the field have criticized the decision, arguing that this set off a race to release technology that gets things wrong, makes things up and could soon be used to rapidly spread disinformation.

Mr. Altman argues that rather than developing and testing the technology entirely behind closed doors before releasing it in full, it is safer to gradually share it so everyone can better understand risks and how to handle them.

He told me that it would be a “very slow takeoff.”

When I asked Mr. Altman if a machine that could do anything the human brain could do would eventually drive the price of human labor to zero, he demurred. He said he could not imagine a world where human intelligence was useless.

If he’s wrong, he thinks he can make it up to humanity.

His grand idea is that OpenAI will capture much of the world’s wealth through the creation of A.G.I. and then redistribute this wealth to the people. In Napa, as we sat chatting beside the lake at the heart of his ranch, he tossed out several figures — $100 billion, $1 trillion, $100 trillion.

If A.G.I. does create all that wealth, he is not sure how the company will redistribute it. Money could mean something very different in this new world.

But as he once told me: “I feel like the A.G.I. can help with that.”

The Tesla Inc. CEO tweeted in February that OpenAI had been founded as an open-source nonprofit “to serve as a counterweight to Google, but now it has become a closed source, maximum-profit company effectively controlled by Microsoft. Not what I intended at all.”

Backers say his brand of social-minded capitalism makes him the ideal person to lead OpenAI. Others, including some who’ve worked for him, say he’s too commercially minded and immersed in Silicon Valley thinking to lead a technological revolution that is already reshaping business and social life. 

In the long run, he said, he wants to set up a global governance structure that would oversee decisions about the future of AI and gradually reduce the power OpenAI’s executive team has over its technology. 

Mr. Altman said he doesn’t necessarily need to be first to develop artificial general intelligence, a world long imagined by researchers and science-fiction writers where software isn’t just good at one specific task like generating text or images but can understand and learn as well or better than a human can. He instead said OpenAI’s ultimate mission is to build AGI, as it’s called, safely.

In its founding charter, OpenAI pledged to abandon its research efforts if another project came close to building AGI before it did. The goal, the company said, was to avoid a race toward building dangerous AI systems fueled by competition and instead prioritize the safety of humanity.

While running Y Combinator, Mr. Altman began to nurse a growing fear that large research labs like DeepMind, purchased by Google in 2014, were creating potentially dangerous AI technologies outside the public eye. Mr. Musk has voiced similar concerns of a dystopian world controlled by powerful AI machines. 

Messrs. Altman and Musk decided it was time to start their own lab. Both were part of a group that pledged $1 billion to the nonprofit, OpenAI Inc. 

OpenAI researchers soon concluded that the most promising path to achieve artificial general intelligence rested in large language models, or computer programs that mimic the way humans read and write. Such models were trained on large volumes of text and required a massive amount of computing power that OpenAI wasn’t equipped to fund as a nonprofit, according to Mr. Altman. 

“We didn’t have a visceral sense of just how expensive this project was going to be,” he said. “We still don’t.”

Tensions also grew with Mr. Musk, who became frustrated with the slow progress and pushed for more control over the organization, people familiar with the matter said. 

OpenAI executives ended up reviving an unusual idea that had been floated earlier in the company’s history: creating a for-profit arm, OpenAI LP, that would report to the nonprofit parent. 

Reid Hoffman, a LinkedIn co-founder who advised OpenAI at the time and later served on the board, said the idea was to attract investors eager to make money from the commercial release of some OpenAI technology, accelerating OpenAI’s progress

“You want to be there first and you want to be setting the norms,” he said. “That’s part of the reason why speed is a moral and ethical thing here.”

The decision further alienated Mr. Musk, the people familiar with the matter said. He parted ways with OpenAI in February 2018. 

Mr. Musk announced his departure in a company all-hands, former employees who attended the meeting said. Mr. Musk explained that he thought he had a better chance at creating artificial general intelligence through Tesla, where he had access to greater resources, they said.

A young researcher questioned whether Mr. Musk had thought through the safety implications, the former employees said. Mr. Musk grew visibly frustrated and called the intern a “jackass,” leaving employees stunned, they said. It was the last time many of them would see Mr. Musk in person.  

Mr. Musk’s departure marked a turning point. Later that year, OpenAI leaders told employees that Mr. Altman was set to lead the company. He formally became CEO and helped complete the creation of the for-profit subsidiary in early 2019.

OpenAI said that it received about $130 million in contributions from the initial $1 billion pledge, but that further donations were no longer needed after the for-profit’s creation. Mr. Musk has tweeted that he donated around $100 million to OpenAI. 

In the meantime, Mr. Altman began hunting for investors. His break came at Allen & Co.’s annual conference in Sun Valley, Idaho in the summer of 2018, where he bumped into Satya Nadella, the Microsoft CEO, on a stairwell and pitched him on OpenAI. Mr. Nadella said he was intrigued. The conversations picked up that winter.

“I remember coming back to the team after and I was like, this is the only partner,” Mr. Altman said. “They get the safety stuff, they get artificial general intelligence. They have the capital, they have the ability to run the compute.”   

Mr. Altman shared the contract with employees as it was being negotiated, hosting all-hands and office hours to allay concerns that the partnership contradicted OpenAI’s initial pledge to develop artificial intelligence outside the corporate world, the former employees said. 

Some employees still saw the deal as a Faustian bargain. 

OpenAI’s lead safety researcher, Dario Amodei, and his lieutenants feared the deal would allow Microsoft to sell products using powerful OpenAI technology before it was put through enough safety testing,

They felt that OpenAI’s technology was far from ready for a large release—let alone with one of the world’s largest software companies—worrying it could malfunction or be misused for harm in ways they couldn’t predict.  

Mr. Amodei also worried the deal would tether OpenAI’s ship to just one company—Microsoft—making it more difficult for OpenAI to stay true to its founding charter’s commitment to assist another project if it got to AGI first, the former employees said.

Microsoft initially invested $1 billion in OpenAI. While the deal gave OpenAI its needed money, it came with a hitch: exclusivity. OpenAI agreed to only use Microsoft’s giant computer servers, via its Azure cloud service, to train its AI models, and to give the tech giant the sole right to license OpenAI’s technology for future products.

In a recent investment deck, Anthropic said it was “committed to large-scale commercialization” to achieve the creation of safe AGI, and that it “fully committed” to a commercial approach in September. The company was founded as an AI safety and research company and said at the time that it might look to create commercial value from its products. 

Mr. Altman “has presided over a 180-degree pivot that seems to me to be only giving lip service to concern for humanity,” he said. 

“The deal completely undermines those tenets to which they secured nonprofit status,” said Gary Marcus, an emeritus professor of psychology and neural science at New York University who co-founded a machine-learning company

The cash turbocharged OpenAI’s progress, giving researchers access to the computing power needed to improve large language models, which were trained on billions of pages of publicly available text. OpenAI soon developed a more powerful language model called GPT-3 and then sold developers access to the technology in June 2020 through packaged lines of code known as application program interfaces, or APIs. 

Mr. Altman and Mr. Amodei clashed again over the release of the API, former employees said. Mr. Amodei wanted a more limited and staged release of the product to help reduce publicity and allow the safety team to conduct more testing on a smaller group of users, former employees said. 

Mr. Amodei left the company a few months later along with several others to found a rival AI lab called Anthropic. “They had a different opinion about how to best get to safe AGI than we did,” Mr. Altman said.

Anthropic has since received more than $300 million from Google this year and released its own AI chatbot called Claude in March, which is also available to developers through an API. 

Mr. Altman disagreed. “The unusual thing about Microsoft as a partner is that it let us keep all the tenets that we think are important to our mission,” he said, including profit caps and the commitment to assist another project if it got to AGI first. 

In the three years after the initial deal, Microsoft invested a total of $3 billion in OpenAI, according to investor documents. 

More than one million users signed up for ChatGPT within five days of its November release, a speed that surprised even Mr. Altman. It followed the company’s introduction of DALL-E 2, which can generate sophisticated images from text prompts.

By February, it had reached 100 million users, according to analysts at UBS, the fastest pace by a consumer app in history to reach that mark.

n’s close associates praise his ability to balance OpenAI’s priorities. No one better navigates between the “Scylla of misplaced idealism” and the “Charybdis of myopic ambition,” Mr. Thiel said. 

Mr. Altman said he delayed the release of the latest version of its model, GPT-4, from last year to March to run additional safety tests. Users had reported some disturbing experiences with the model, integrated into Bing, where the software hallucinated—meaning it made up answers to questions it didn’t know. It issued ominous warnings and made threats. 

“The way to get it right is to have people engage with it, explore these systems, study them, to learn how to make them safe,” Mr. Altman said.

After Microsoft’s initial investment is paid back, it would capture 49% of OpenAI’s profits until the profit cap, up from 21% under prior arrangements, the documents show. OpenAI Inc., the nonprofit parent, would get the rest.

He has put almost all his liquid wealth in recent years in two companies. He has put $375 million into Helion Energy, which is seeking to create carbon-free energy from nuclear fusion and is close to creating “legitimate net-gain energy in a real demo,” Mr. Altman said.

He has also put $180 million into Retro, which aims to add 10 years to the human lifespan through “cellular reprogramming, plasma-inspired therapeutics and autophagy,” or the reuse of old and damaged cell parts, according to the company. 

He noted how much easier these problems are, morally, than AI. “If you’re making nuclear fusion, it’s all upside. It’s just good,” he said. “If you’re making AI, it is potentially very good, potentially very terrible.” 

The eerily humanlike chatbot, which went viral last year by generating its own answers to prompts, blindsided E.U. policymakers. The type of A.I. that powered ChatGPT was not mentioned in the draft law and was not a major focus of discussions about the policy. Lawmakers and their aides peppered one another with calls and texts to address the gap, as tech executives warned that overly aggressive regulations could put Europe at an economic disadvantage.

Even now, E.U. lawmakers are arguing over what to do, putting the law at risk. “We will always be lagging behind the speed of technology,” said Svenja Hahn, a member of the European Parliament who was involved in writing the A.I. law.

Lawmakers and regulators in Brussels, in Washington and elsewhere are losing a battle to regulate A.I. and are racing to catch up, as concerns grow that the powerful technology will automate away jobs, turbocharge the spread of disinformation and eventually develop its own kind of intelligence.

Nations have moved swiftly to tackle A.I.’s potential perils, but European officials have been caught off guard by the technology’s evolution, while U.S. lawmakers openly concede that they barely understand how it works.

The absence of rules has left a vacuum. Google, Meta, Microsoft and OpenAI, which makes ChatGPT, have been left to police themselves as they race to create and profit from advanced A.I. systems

At the root of the fragmented actions is a fundamental mismatch. A.I. systems are advancing so rapidly and unpredictably that lawmakers and regulators can’t keep pace

That gap has been compounded by an A.I. knowledge deficit in governments, labyrinthine bureaucracies and fears that too many rules may inadvertently limit the technology’s benefits.

Even in Europe, perhaps the world’s most aggressive tech regulator, A.I. has befuddled policymakers.

The European Union has plowed ahead with its new law, the A.I. Act, despite disputes over how to handle the makers of the latest A.I. systems.

The result has been a sprawl of responses. President Biden issued an executive order in October about A.I.’s national security effects as lawmakers debate what, if any, measures to pass. Japan is drafting nonbinding guidelines for the technology, while China has imposed restrictions on certain types of A.I. Britain has said existing laws are adequate for regulating the technology. Saudi Arabia and the United Arab Emirates are pouring government money into A.I. research.

A final agreement, expected as soon as Wednesday, could restrict certain risky uses of the technology and create transparency requirements about how the underlying systems work. But even if it passes, it is not expected to take effect for at least 18 months — a lifetime in A.I. development — and how it will be enforced is unclear.

Many companies, preferring nonbinding codes of conduct that provide latitude to speed up development, are lobbying to soften proposed regulations and pitting governments against one another.

“No one, not even the creators of these systems, know what they will be able to do,” said Matt Clifford, an adviser to Prime Minister Rishi Sunak of Britain, who presided over an A.I. Safety Summit last month with 28 countries. “The urgency comes from there being a real question of whether governments are equipped to deal with and mitigate the risks.”

Europe takes the lead

In mid-2018, 52 academics, computer scientists and lawyers met at the Crowne Plaza hotel in Brussels to discuss artificial intelligence. E.U. officials had selected them to provide advice about the technology, which was drawing attention for powering driverless cars and facial recognition systems.

as they discussed A.I.’s possible effects — including the threat of facial recognition technology to people’s privacy — they recognized “there were all these legal gaps, and what happens if people don’t follow those guidelines?”

In 2019, the group published a 52-page report with 33 recommendations, including more oversight of A.I. tools that could harm individuals and society.

By October, the governments of France, Germany and Italy, the three largest E.U. economies, had come out against strict regulation of general purpose A.I. models for fear of hindering their domestic tech start-ups. Others in the European Parliament said the law would be toothless without addressing the technology. Divisions over the use of facial recognition technology also persisted.

So when the A.I. Act was unveiled in 2021, it concentrated on “high risk” uses of the technology, including in law enforcement, school admissions and hiring. It largely avoided regulating the A.I. models that powered them unless listed as dangerous

“They sent me a draft, and I sent them back 20 pages of comments,” said Stuart Russell, a computer science professor at the University of California, Berkeley, who advised the European Commission. “Anything not on their list of high-risk applications would not count, and the list excluded ChatGPT and most A.I. systems.”

E.U. leaders were undeterred.“Europe may not have been the leader in the last wave of digitalization, but it has it all to lead the next one,” Ms. Vestager said when she introduced the policy at a news conference in Brussels.

In 2020, European policymakers decided that the best approach was to focus on how A.I. was used and not the underlying technology. A.I. was not inherently good or bad, they said — it depended on how it was applied.

Nineteen months later, ChatGPT arrived.

The Washington game

Lacking tech expertise, lawmakers are increasingly relying on Anthropic, Microsoft, OpenAI, Google and other A.I. makers to explain how it works and to help create rules.

“We’re not experts,” said Representative Ted Lieu, Democrat of California, who hosted Sam Altman, OpenAI’s chief executive, and more than 50 lawmakers at a dinner in Washington in May. “It’s important to be humble.”

Tech companies have seized their advantage. In the first half of the year, many of Microsoft’s and Google’s combined 169 lobbyists met with lawmakers and the White House to discuss A.I. legislation, according to lobbying disclosures. OpenAI registered its first three lobbyists and a tech lobbying group unveiled a $25 million campaign to promote A.I.’s benefits this year.

In that same period, Mr. Altman met with more than 100 members of Congress, including former Speaker Kevin McCarthy, Republican of California, and the Senate leader, Chuck Schumer, Democrat of New York. After testifying in Congress in May, Mr. Altman embarked on a 17-city global tour, meeting world leaders including President Emmanuel Macron of France, Mr. Sunak and Prime Minister Narendra Modi of India.

, the White House announced that the four companies had agreed to voluntary commitments on A.I. safety, including testing their systems through third-party overseers — which most of the companies were already doing.

“It was brilliant,” Mr. Smith said. “Instead of people in government coming up with ideas that might have been impractical, they said, ‘Show us what you think you can do and we’ll push you to do more.’”

In a statement, Ms. Raimondo said the federal government would keep working with companies so “America continues to lead the world in responsible A.I. innovation.”

Over the summer, the Federal Trade Commission opened an investigation into OpenAI and how it handles user data. Lawmakers continued welcoming tech executives.

In September, Mr. Schumer was the host of Elon Musk, Mark Zuckerberg of Meta, Sundar Pichai of Google, Satya Nadella of Microsoft and Mr. Altman at a closed-door meeting with lawmakers in Washington to discuss A.I. rules. Mr. Musk warned of A.I.’s “civilizational” risks, while Mr. Altman proclaimed that A.I. could solve global problems such as poverty.

A.I. companies are playing governments off one another. In Europe, industry groups have warned that regulations could put the European Union behind the United States. In Washington, tech companies have cautioned that China might pull ahead.

In May, Ms. Vestager, Ms. Raimondo and Antony J. Blinken, the U.S. secretary of state, met in Lulea, Sweden, to discuss cooperating on digital policy.

“China is way better at this stuff than you imagine,” Mr. Clark of Anthropic told members of Congress in January.

After two days of talks, Ms. Vestager announced that Europe and the United States would release a shared code of conduct for safeguarding A.I. “within weeks.” She messaged colleagues in Brussels asking them to share her social media post about the pact, which she called a “huge step in a race we can’t afford to lose.”

Months later, no shared code of conduct had appeared. The United States instead announced A.I. guidelines of its own.

Little progress has been made internationally on A.I. With countries mired in economic competition and geopolitical distrust, many are setting their own rules for the borderless technology.

Yet “weak regulation in another country will affect you,” said Rajeev Chandrasekhar, India’s technology minister, noting that a lack of rules around American social media companies led to a wave of global disinformation.

“Most of the countries impacted by those technologies were never at the table when policies were set,” he said. “A.I will be several factors more difficult to manage.”

Even among allies, the issue has been divisive. At the meeting in Sweden between E.U. and U.S. officials, Mr. Blinken criticized Europe for moving forward with A.I. regulations that could harm American companies, one attendee said. Thierry Breton, a European commissioner, shot back that the United States could not dictate European policy, the person said.

Some policymakers said they hoped for progress at an A.I. safety summit that Britain held last month at Bletchley Park, where the mathematician Alan Turing helped crack the Enigma code used by the Nazis. The gathering featured Vice President Kamala Harris; Wu Zhaohui, China’s vice minister of science and technology; Mr. Musk; and others.

The upshot was a 12-paragraph statement describing A.I.’s “transformative” potential and “catastrophic” risk of misuse. Attendees agreed to meet again next year.

The talks, in the end, produced a deal to keep talking.

This artificial intelligence breakthrough, he argues, is being driven by cheap parallel computation technologies, big data collection and better algorithms. The upshot is clear, “The business plans of the next 10,000 start-ups are easy to forecast: Take X and add A.I.”

in 2001, the top 10 websites accounted for 31 percent of all U.S. page views, but, by 2010, they accounted for 75 percent of them.

The Internet has created a long tail, but almost all the revenue and power is among the small elite at the head.

Advances in artificial intelligence will accelerate this centralizing trend. That’s because A.I. companies will be able to reap the rewards of network effects. The bigger their network and the more data they collect, the more effective and attractive they become.

As a result, our A.I. future is likely to be ruled by an oligarchy of two or three large, general-purpose cloud-based commercial intelligences.”

engineers at a few gigantic companies will have vast-though-hidden power to shape how data are collected and framed, to harvest huge amounts of information, to build the frameworks through which the rest of us make decisions and to steer our choices. If you think this power will be used for entirely benign ends, then you have not read enough history.

The second implication is philosophical. A.I. will redefine what it means to be human. Our identity as humans is shaped by what machines and other animals can’t do

For the last few centuries, reason was seen as the ultimate human faculty. But now machines are better at many of the tasks we associate with thinking — like playing chess, winning at Jeopardy, and doing math.

On the other hand, machines cannot beat us at the things we do without conscious thinking: developing tastes and affections, mimicking each other and building emotional attachments, experiencing imaginative breakthroughs, forming moral sentiments.

In the age of smart machines, we’re not human because we have big brains. We’re human because we have social skills, emotional capacities and moral intuitions.

I could paint two divergent A.I. futures, one deeply humanistic, and one soullessly utilitarian.

In the humanistic one, machines liberate us from mental drudgery so we can focus on higher and happier things. In this future, differences in innate I.Q. are less important. Everybody has Google on their phones so having a great memory or the ability to calculate with big numbers doesn’t help as much.

In this future, there is increasing emphasis on personal and moral faculties: being likable, industrious, trustworthy and affectionate. People are evaluated more on these traits, which supplement machine thinking, and not the rote ones that duplicate it

In the cold, utilitarian future, on the other hand, people become less idiosyncratic. If the choice architecture behind many decisions is based on big data from vast crowds, everybody follows the prompts and chooses to be like each other. The machine prompts us to consume what is popular, the things that are easy and mentally undemanding.

The key issue is not “human-competitive” intelligence (as the open letter puts it); it’s what happens after AI gets to smarter-than-human intelligence. Key thresholds there may not be obvious, we definitely can’t calculate in advance what happens when, and it currently seems imaginable that a research lab would cross critical lines without noticing.

Many researchers steeped in these issues, including myself, expect that the most likely result of building a superhumanly smart AI, under anything remotely like the current circumstances, is that literally everyone on Earth will die. Not as in “maybe possibly some remote chance,” but as in “that is the obvious thing that would happen.”

It’s not that you can’t, in principle, survive creating something much smarter than you; it’s that it would require precision and preparation and new scientific insights, and probably not having AI systems composed of giant inscrutable arrays of fractional numbers.

Absent that caring, we get “the AI does not love you, nor does it hate you, and you are made of atoms it can use for something else.”

Without that precision and preparation, the most likely outcome is AI that does not do what we want, and does not care for us nor for sentient life in general. That kind of caring is something that could in principle be imbued into an AI but we are not ready and do not currently know how.

The likely result of humanity facing down an opposed superhuman intelligence is a total loss

To visualize a hostile superhuman AI, don’t imagine a lifeless book-smart thinker dwelling inside the internet and sending ill-intentioned emails. Visualize an entire alien civilization, thinking at millions of times human speeds, initially confined to computers—in a world of creatures that are, from its perspective, very stupid and very slow. A sufficiently intelligent AI won’t stay confined to computers for long. In today’s world you can email DNA strings to laboratories that will produce proteins on demand, allowing an AI initially confined to the internet to build artificial life forms or bootstrap straight to postbiological molecular manufacturing.

There’s no proposed plan for how we could do any such thing and survive. OpenAI’s openly declared intention is to make some future AI do our AI alignment homework. Just hearing that this is the plan ought to be enough to get any sensible person to panic. The other leading AI lab, DeepMind, has no plan at all.

An aside: None of this danger depends on whether or not AIs are or can be conscious; it’s intrinsic to the notion of powerful cognitive systems that optimize hard and calculate outputs that meet sufficiently complicated outcome criteria.

I didn’t also mention that we have no idea how to determine whether AI systems are aware of themselves—since we have no idea how to decode anything that goes on in the giant inscrutable arrays—and therefore we may at some point inadvertently create digital minds which are truly conscious and ought to have rights and shouldn’t be owned.

I refrained from signing because I think the letter is understating the seriousness of the situation and asking for too little to solve it.

the thing about trying this with superhuman intelligence is that if you get that wrong on the first try, you do not get to learn from your mistakes, because you are dead. Humanity does not learn from the mistake and dust itself off and try again, as in other challenges we’ve overcome in our history, because we are all gone.

If we held anything in the nascent field of Artificial General Intelligence to the lesser standards of engineering rigor that apply to a bridge meant to carry a couple of thousand cars, the entire field would be shut down tomorrow.

We are not on course to be prepared in any reasonable time window. There is no plan. Progress in AI capabilities is running vastly, vastly ahead of progress in AI alignment or even progress in understanding what the hell is going on inside those systems

Many researchers working on these systems think that we’re plunging toward a catastrophe, with more of them daring to say it in private than in public; but they think that they can’t unilaterally stop the forward plunge, that others will go on even if they personally quit their jobs.

This is a stupid state of affairs, and an undignified way for Earth to die, and the rest of humanity ought to step in at this point and help the industry solve its collective action problem.

When the insider conversation is about the grief of seeing your daughter lose her first tooth, and thinking she’s not going to get a chance to grow up, I believe we are past the point of playing political chess about a six-month moratorium.

The moratorium on new large training runs needs to be indefinite and worldwide. There can be no exceptions, including for governments or militaries. If the policy starts with the U.S., then China needs to see that the U.S. is not seeking an advantage but rather trying to prevent a horrifically dangerous technology which can have no true owner and which will kill everyone in the U.S. and in China and on Earth

Here’s what would actually need to be done:

Shut down all the large GPU clusters (the large computer farms where the most powerful AIs are refined). Shut down all the large training runs

Put a ceiling on how much computing power anyone is allowed to use in training an AI system, and move it downward over the coming years to compensate for more efficient training algorithm

Track all GPUs sold. If intelligence says that a country outside the agreement is building a GPU cluster, be less scared of a shooting conflict between nations than of the moratorium being violated; be willing to destroy a rogue datacenter by airstrike.

Frame nothing as a conflict between national interests, have it clear that anyone talking of arms races is a fool

Make it explicit in international diplomacy that preventing AI extinction scenarios is considered a priority above preventing a full nuclear exchange, and that allied nuclear countries are willing to run some risk of nuclear exchange if that’s what it takes to reduce the risk of large AI training runs.

when your policy ask is that large, the only way it goes through is if policymakers realize that if they conduct business as usual, and do what’s politically easy, that means their own kids are going to die too.

Two sources said that during the early weeks of the war they were permitted to kill 15 or 20 civilians during airstrikes on low-ranking militants. Attacks on such targets were typically carried out using unguided munitions known as “dumb bombs”, the sources said, destroying entire homes and killing all their occupants.

“You don’t want to waste expensive bombs on unimportant people – it’s very expensive for the country and there’s a shortage [of those bombs],” one intelligence officer said. Another said the principal question they were faced with was whether the “collateral damage” to civilians allowed for an attack.

“Because we usually carried out the attacks with dumb bombs, and that meant literally dropping the whole house on its occupants. But even if an attack is averted, you don’t care – you immediately move on to the next target. Because of the system, the targets never end. You have another 36,000 waiting.”

ccording to conflict experts, if Israel has been using dumb bombs to flatten the homes of thousands of Palestinians who were linked, with the assistance of AI, to militant groups in Gaza, that could help explain the shockingly high death toll in the war.

Details about the specific kinds of data used to train Lavender’s algorithm, or how the programme reached its conclusions, are not included in the accounts published by +972 or Local Call. However, the sources said that during the first few weeks of the war, Unit 8200 refined Lavender’s algorithm and tweaked its search parameters.

Responding to the publication of the testimonies in +972 and Local Call, the IDF said in a statement that its operations were carried out in accordance with the rules of proportionality under international law. It said dumb bombs are “standard weaponry” that are used by IDF pilots in a manner that ensures “a high level of precision”.

“The IDF does not use an artificial intelligence system that identifies terrorist operatives or tries to predict whether a person is a terrorist,” it added. “Information systems are merely tools for analysts in the target identification process.”

In earlier military operations conducted by the IDF, producing human targets was often a more labour-intensive process. Multiple sources who described target development in previous wars to the Guardian, said the decision to “incriminate” an individual, or identify them as a legitimate target, would be discussed and then signed off by a legal adviser.

n the weeks and months after 7 October, this model for approving strikes on human targets was dramatically accelerated, according to the sources. As the IDF’s bombardment of Gaza intensified, they said, commanders demanded a continuous pipeline of targets.

“We were constantly being pressured: ‘Bring us more targets.’ They really shouted at us,” said one intelligence officer. “We were told: now we have to fuck up Hamas, no matter what the cost. Whatever you can, you bomb.”

Lavender was developed by the Israel Defense Forces’ elite intelligence division, Unit 8200, which is comparable to the US’s National Security Agency or GCHQ in the UK.

After randomly sampling and cross-checking its predictions, the unit concluded Lavender had achieved a 90% accuracy rate, the sources said, leading the IDF to approve its sweeping use as a target recommendation tool.

Lavender created a database of tens of thousands of individuals who were marked as predominantly low-ranking members of Hamas’s military wing, they added. This was used alongside another AI-based decision support system, called the Gospel, which recommended buildings and structures as targets rather than individuals.

The accounts include first-hand testimony of how intelligence officers worked with Lavender and how the reach of its dragnet could be adjusted. “At its peak, the system managed to generate 37,000 people as potential human targets,” one of the sources said. “But the numbers changed all the time, because it depends on where you set the bar of what a Hamas operative is.”

broadly, and then the machine started bringing us all kinds of civil defence personnel, police officers, on whom it would be a shame to waste bombs. They help the Hamas government, but they don’t really endanger soldiers.”

Before the war, US and Israeli estimated membership of Hamas’s military wing at approximately 25-30,000 people.

there was a decision to treat Palestinian men linked to Hamas’s military wing as potential targets, regardless of their rank or importance.

According to +972 and Local Call, the IDF judged it permissible to kill more than 100 civilians in attacks on a top-ranking Hamas officials. “We had a calculation for how many [civilians could be killed] for the brigade commander, how many [civilians] for a battalion commander, and so on,” one source said.

Another source, who justified the use of Lavender to help identify low-ranking targets, said that “when it comes to a junior militant, you don’t want to invest manpower and time in it”. They said that in wartime there was insufficient time to carefully “incriminate every target”

So you’re willing to take the margin of error of using artificial intelligence, risking collateral damage and civilians dying, and risking attacking by mistake, and to live with it,” they added.

When it came to targeting low-ranking Hamas and PIJ suspects, they said, the preference was to attack when they were believed to be at home. “We were not interested in killing [Hamas] operatives only when they were in a military building or engaged in a military activity,” one said. “It’s much easier to bomb a family’s home. The system is built to look for them in these situations.”

Such a strategy risked higher numbers of civilian casualties, and the sources said the IDF imposed pre-authorised limits on the number of civilians it deemed acceptable to kill in a strike aimed at a single Hamas militant. The ratio was said to have changed over time, and varied according to the seniority of the target.

The IDF’s targeting processes in the most intensive phase of the bombardment were also relaxed, they said. “There was a completely permissive policy regarding the casualties of [bombing] operations,” one source said. “A policy so permissive that in my opinion it had an element of revenge.”

“There were regulations, but they were just very lenient,” another added. “We’ve killed people with collateral damage in the high double digits, if not low triple digits. These are things that haven’t happened before.” There appears to have been significant fluctuations in the figure that military commanders would tolerate at different stages of the war

One source said that the limit on permitted civilian casualties “went up and down” over time, and at one point was as low as five. During the first week of the conflict, the source said, permission was given to kill 15 non-combatants to take out junior militants in Gaza

at one stage earlier in the war they were authorised to kill up to “20 uninvolved civilians” for a single operative, regardless of their rank, military importance, or age.

“It’s not just that you can kill any person who is a Hamas soldier, which is clearly permitted and legitimate in terms of international law,” they said. “But they directly tell you: ‘You are allowed to kill them along with many civilians.’ … In practice, the proportionality criterion did not exist.”

Experts in international humanitarian law who spoke to the Guardian expressed alarm at accounts of the IDF accepting and pre-authorising collateral damage ratios as high as 20 civilians, particularly for lower-ranking militants. They said militaries must assess proportionality for each individual strike.

An international law expert at the US state department said they had “never remotely heard of a one to 15 ratio being deemed acceptable, especially for lower-level combatants. There’s a lot of leeway, but that strikes me as extreme”.

Sarah Harrison, a former lawyer at the US Department of Defense, now an analyst at Crisis Group, said: “While there may be certain occasions where 15 collateral civilian deaths could be proportionate, there are other times where it definitely wouldn’t be. You can’t just set a tolerable number for a category of targets and say that it’ll be lawfully proportionate in each case.”

Whatever the legal or moral justification for Israel’s bombing strategy, some of its intelligence officers appear now to be questioning the approach set by their commanders. “No one thought about what to do afterward, when the war is over, or how it will be possible to live in Gaza,” one said.

Another said that after the 7 October attacks by Hamas, the atmosphere in the IDF was “painful and vindictive”. “There was a dissonance: on the one hand, people here were frustrated that we were not attacking enough. On the other hand, you see at the end of the day that another thousand Gazans have died, most of them civilians.”

With greater accuracy, for example, marketers can comb large databases of consumer behavior to get more precise information on buying habits. And improvements in facial recognition are likely to make surveillance technology cheaper and more commonplace.

Modern artificial neural networks are composed of an array of software components, divided into inputs, hidden layers and outputs. The arrays can be “trained” by repeated exposures to recognize patterns like images or sounds.

Authoritarian governments have created seemingly realistic news broadcasters to advance their political goals

Experts fear the technology could hasten an erosion of trust in media, in government and in society. If any image can be manufactured — and manipulated — how can we believe anything we see?

“The tools are going to get better, they’re going to get cheaper, and there will come a day when nothing you see on the internet can be believed,” said Wasim Khaled, chief executive of Blackbird.AI, a company that helps clients fight disinformation.

Artificial intelligence allows virtually anyone to create complex artworks, like those now on exhibit at the Gagosian art gallery in New York, or lifelike images that blur the line between what is real and what is fiction. Plug in a text description, and the technology can produce a related image — no special skills required.

Midjourney’s images, he said, were able to pass muster in facial-recognition programs that Bellingcat uses to verify identities, typically of Russians who have committed crimes or other abuses. It’s not hard to imagine governments or other nefarious actors manufacturing images to harass or discredit their enemies.

In February, Getty accused Stability AI of illegally copying more than 12 million Getty photos, along with captions and metadata, to train the software behind its Stable Diffusion tool. In its lawsuit, Getty argued that Stable Diffusion diluted the value of the Getty watermark by incorporating it into images that ranged “from the bizarre to the grotesque.”

Getty’s lawsuit reflects concerns raised by many individual artists — that A.I. companies are becoming a competitive threat by copying content they do not have permission to use.

Trademark violations have also become a concern: Artificially generated images have replicated NBC’s peacock logo, though with unintelligible letters, and shown Coca-Cola’s familiar curvy logo with extra O’s looped into the name.

The threat to photographers is fast outpacing the development of legal protections, said Mickey H. Osterreicher, general counsel for the National Press Photographers Association

Newsrooms will increasingly struggle to authenticate conten

Social media users are ignoring labels that clearly identify images as artificially generated, choosing to believe they are real photographs, he said.

The video explained that the deepfake had been created, with Ms. Schick’s consent, by the Dutch company Revel.ai and Truepic, a California company that is exploring broader digital content verification

The companies described their video, which features a stamp identifying it as computer-generated, as the “first digitally transparent deepfake.” The data is cryptographically sealed into the file; tampering with the image breaks the digital signature and prevents the credentials from appearing when using trusted software.

The companies hope the badge, which will come with a fee for commercial clients, will be adopted by other content creators to help create a standard of trust involving A.I. images.

“The scale of this problem is going to accelerate so rapidly that it’s going to drive consumer education very quickly,” said Jeff McGregor, chief executive of Truepic

Adobe unveiled its own image-generating product, Firefly, which will be trained using only images that were licensed or from its own stock or no longer under copyright. Dana Rao, the company’s chief trust officer, said on its website that the tool would automatically add content credentials — “like a nutrition label for imaging” — that identified how an image had been made. Adobe said it also planned to compensate contributors.

“Once humans develop artificial intelligence, it will take off on its own and redesign itself at an ever-increasing rate,” Hawking warned in recent months. “Humans, who are limited by slow biological evolution, couldn't compete and would be superseded.”