Is it possible the latest artificial intelligence hype cycle is just another scam? Sci-fi author Charles Stross, someone who’s thought deeply and written at length about artificial intelligence, certainly seems to think so:
The thing I find most suspicious/fishy/smelly about the current hype surrounding Stable Diffusion, ChatGPT, and other AI applications is that it is almost exactly six months since the bottom dropped out of the cryptocurrency scam bubble.
Suspicious, indeed. I’ll take it one step further. What if this AI cycle, in particular the artificial general intelligence (AGI) discourse/panic kicked off by Sam Altman’s latest pronouncement, is a signal we’re entering a new, more dangerous era of tech hype? Whereas past tech promises at least attempted to apply technology to societal problems, AGI, like crypto/the metaverse before it, are doing something else: promising to side-step society entirely.
Until I come up with a better term, I’m calling this escalation the Big Promise. As I discussed a few weeks ago, information technology companies have always made grandiose claims about what their technology will eventually accomplish as a way of justifying their stratospheric valuations and powerful positions in society, and distracting employees from the fact that they’re mostly in the business of selling gizmos and ads. But in past hype eras, these failed promises — self-driving cars, AI assistants, autonomous robots, universal global internet, solving death, etc. — were at least relatively focused on solving specific societal problems.
As most of those smaller promises continue to sputter, tech is now facing a problem it has never really encountered before. Unlike during the dot-com bubble, tech has now come to dominate the global economy. Seven out of the 10 most valuable companies are tech firms. In 1999, that number was 3-4, depending on if you count Nokia as telephony or tech. The economic stakes are much higher today.
The stakes for societal backlash are much higher, too. Yes, companies in the dot-com era made lots of crazy promises, but because the information technology revolution was still in its infancy — less than half of American households owned a computer, and just a quarter had an internet connection in 1999 — far fewer people knew about those promises, much less believed that any of them would come true. Today, near-universal internet penetration is creating an accelerating, society-wide tech backlash just as investment is beginning to dry up due to the global economy facing its first high-interest environment in a generation. In other words, the conditions are set for something of a big tech calamity. If the majority of society ever stops believing that tech can eventually solve all its problems, which is now on the table, those sky-high valuations will drop precipitously and a lot of people — in and out of tech — will lose a lot of money and jobs with no VC dollars available to act as a safety net.
So how do you keep the dream alive and the dollars flowing even now, after decades of failure to live up to your promises? Easy. You pretend the problem isn’t technology, but society itself. And you escalate the promise from fixing society’s problems by replacing humans with computers to replacing society itself. With, you guessed it, computers.
Crypto/web3/the metaverse is a perfect example of the Big Promise. Though very nebulously defined, the concept is essentially a technology stack for allowing the powerful to create their own bubbles of neoliberalism to play in, free of the physical world and its annoying existential threats, politics, people we don’t like, and regulated fiat currencies. Its proponents literally argue that we should build new societies where only events that can be expressed on the blockchain are valuable. No need for self-driving cars if you can afford the gas fees to teleport to our friend’s virtual party, I guess.
Luckily, just as that fabulous idea started to run out of expensive pixelated steam, another one rose to take its place. Like the blockchain before it, ChatGPT, Stable Diffusion, and so on are interesting technologies that will probably prove useful for something, eventually? But, as investor William Eden argues, if you look beyond today’s demos, you’ll realize that they’re both much more expensive than traditional approaches to information retrieval/creation and also not very reliable. As a result, they can’t actually function in environments where the stakes, and thus the value, are high. So they need the Big Promise to justify all the money being poured into them.
Enter Artificial General Intelligence. Like web3/the metaverse before it, AGI is another fuzzy concept that can mean just about anything you want it to, but is generally understood to mean building a world-saving/destroying superintelligence that runs everything for us. Or, as Altman puts it: “If AGI is successfully created, this technology could help us elevate humanity by increasing abundance, turbocharging the global economy, and aiding in the discovery of new scientific knowledge that changes the limits of possibility.”
In other words, society has failed to solve its own problems, so give us enough money and power and eventually we’ll create a giant brain that will figure it all out for us. I know that sounds a little glib, but that really is the core promise. Here’s Deep Mind, Alphabet’s AI company, on AGI and climate change:
Advances in AGI research will supercharge society’s ability to tackle and manage climate change - not least because of its urgency but also due to its complex and multifaceted nature.
Big if true, and to be fair, AI has helped scientists better model weather patterns and changes and we should probably keep working on that. But it’s also true that we already have all the technology we need to solve climate change. We just aren’t yet capable of convincing society to implement it fast enough.
AGI conveniently offers a solution for this problem, too: eventually we’ll just let the benevolent super-intelligence we create run the world for us. Again, I’m not really exaggerating. Back to Altman: “Successfully transitioning to a world with superintelligence is perhaps the most important—and hopeful, and scary—project in human history.”
There’s just one problem. How could anyone actually believe that a man who has already decided that he has earned the power to determine the destiny of the entire world despite acknowledging the literally apocalyptic consequences — “AGI would also come with serious risk of misuse, drastic accidents, and societal disruption. Because the upside of AGI is so great, we do not believe it is possible or desirable for society to stop its development forever; instead, society and the developers of AGI have to figure out how to get it right” — will listen to an AGI that tells him something he doesn’t want to hear?
Here’s a thought experiment by way of the Register (the best/only? tech tabloid) that neatly answers that question:
Imagine an AGI system that advises taxing billionaires at a rate of 95 percent and redistributing their wealth for the benefit of humanity. Will it ever be hooked into the banking system to effect its recommended changes?
Unfortunately, as tech leaders may have learned from Donald Trump’s Big Lie, if you are powerful enough and loud enough, you can convince a significant number of people to believe even the most outlandish falsehood despite all evidence to the contrary — precisely because they can’t believe that a rich, powerful person would be crazy enough to make up something so farfetched.
Thus, first with crypto and now with AGI, we have entered the era of tech’s Big Promise: despite all the harm it has (if I’m being charitable, unintentionally) wrought on society, despite helping to drive an unprecedented amount of worldwide inequality, despite a litany of past failures to solve concrete societal problems, if we just invest even more in big tech, it will deliver a solution to society itself in the form of a literal benevolent God-computer (but don’t worry, we can control it).
This has been a promise of AI for decades; unlike crypto the path there is relatively straight and paved with good computing, though it’s never been clear whether it would arrive anywhere. Now that destinations are starting to appear, it’s unkind to compare it to scams.
But it will definitely to first order amplify current inequalities, and will not in itself (without people successfully fighting for a societal transformation) bring about any sort of revolution. There is no more reason to believe in a b. G-c. than to believe in god-emperor Clippy.
I’m not comparing AI to scams. I am comparing the latest hype cycle driving investment in it to scams. Specifically, I am casting doubt on the claim that the current cycle will actually lead to any of those destination — one of which, made by the current leader of the field, is literally a God-computer — and spending a lot of time defending that view.
It’s worth distinguishing “general intelligence” that is broadly useful and can be part of everyday interactions + problem solving as well as introspection and updating (what I’d call AGI), from accelerating change (a singular* transition to a new equilibrium), from the eschatology of superbeings and end times.
I agree — but that’s also kind of the point. No one is making the distinction, at least not in the hype cycles.
unlike the three examples in the previous ¶, these are dominating existing subfields (ad copy, stock images, video compression, video production, &c) with faster, lower-power, customizable alternatives.
I’m skeptical that this is true. See Eden’s thread, specifically: