(Non-fiction. In fact, worse than fiction: opinion.)
Fear not. This is not a rant about AI, either for or against. It’s bad enough being in the midst of the explosion of a new technology about which we cannot fathom the full implications, without having to also put up with a bunch of randos frothing and fuming over it. The level of discussion of AI is an abysmal indictment on the human intellect. Commentators are either in a panic, or they are swooning; a sober, informative account appears to be out of the question1.
The gibberish is understandable. I suspect there were equally laughable warnings and predictions from the time of the advent of the computer, or the internal combustion engine, or the printing press. And if you are a professional pontificator, there is money to be made from wading into the frenzy. But if you are a sensible bystander, history assures you that the hype and the hysteria are entirely ignorable. They tell you nothing useful.
That said… one interesting and hopeful thought occurred to me recently.
I realised that the vast new AI industry is the best thing that could have happened to the study of human consciousness.
Before you swipe away in disgust: this has nothing to do with ludicrous claims that large language models are conscious.
Ok, sorry, it has everything to do with such claims — although the claims are ludicrous nonetheless. There is a very small chance that LLMs are conscious, and it is larger than the chance that my teapot is conscious. However, the claims are ludicrous because we do not have a scientific definition of consciousness, and therefore no means to robustly verify that anything is conscious.
Except of course human beings, because that’s what consciousness is: a thing that human beings experience. (We love a tautology over here in Science Land.) Erik Hoel has written an entire article on the definition of consciousness, where he gets amusingly irritated by people who claim that consciousness is entirely undefined, and wants to clarify for anyone who is listening that of course we all know what we mean by consciousness, it’s just a scientific definition that we’re missing. A “scientific definition” means that you have a sufficiently informative and agreed-on model of consciousness that you can, among other things, reliably determine when it does and does not exist.
Hoel has recently written another article, which prompted the grand think-piece you are currently reading, where he argues that for a long time the study of consciousness was merely a pleasant pastime, but no more. Now the problem is urgent — because of AI. Which brings us to the happy hopeful thought that I want to talk about here2.
The booming AI industry is going to supercharge the study of consciousness. AI is now a trillion-dollar industry. That’s not a joke. A google AI summary confirmed it. A trillion dollar industry. The companies involved are spending astronomical sums of money not only on developing new models, but on research into AI methods, including how they work and what they do and what their potential is. At the headiest echelons of this work is the question of what it means to have intelligence — and what it means to be conscious. You can be sure that a flood of money is going to wash over consciousness research.
This reminds me of a book I read, about a decade ago, on the history of quantum mechanics3. It recounted Planck’s famous discovery of the quantum idea, but in a quite different way to how I learned it in physics classes.
In physics classes we were told about the “blackbody spectrum”, which I’m not going to pretend is interesting4. The point is that objects give off light when they are hot, like red-hot pokers, and the amount of light they give off at different frequencies depends on how hot they are. That’s all very well, but at the end of the 19th century no-one could explain the specific dependence on temperature that was measured in experiments. Then Planck came up with a model where he pretended that energy couldn’t vary continuously, but only in discrete amounts, like the difference between being able to turn a knob freely and having to click from one setting to the next. At the time it was just a convenient trick to get the right answer, but it turned out to revolutionise our understanding of nature. Hurrah!
That was the usual story. The history book I read included some extra context. Planck wasn’t just fiddling around with an amusing problem that he and his other physics chums had hit upon. Science history loves a story of a lone genius stumbling upon a brilliant discovery while just going about their usual business of being absent-mindedly clever. But it wasn’t like that!
At the end of the 19th century electricity had become big business. One part of the business, which may sound quaint in the midst of a discussion about trillions of dollars being spent on AI, but was absolutely huge at the time, was light bulbs. Light bulbs were a major new industry, and making them as efficient as possible was a major challenge. Unfortunately, there was this pesky problem that no-one could explain the spectrum of light from light-bulb filaments, and this was an impassable obstacle to better efficiency. It was so serious that the German government, which was hoping new technologies would give it an edge over its industrial rivals, funded an entire institute devoted to industrial technology, and it included a research unit on blackbody radiation.
So, in this telling, the discovery of quantum mechanics was entirely the result of industrial pressure5. There was money to be made in explaining blackbody radiation, and the search lead to a surprising discovery. As much as we deep-thinking scientists like to imagine ourselves as pure of soul and divorced from petty human concerns, our research directions are massively (one could say entirely) determined by where the money points.
Right now the money points very definitively at AI, and AI boffins are all in a tizz about consciousness. As Hoel somewhat glumly points out, up until now, despite a few popular books and some buzz among the chattering classes, consciousness research has received extremely little funding. He estimates a few million dollars over a decade. That’s surely an underestimate: two million dollars would barely fund one postdoc for a decade. Never mind. If consciousness research attracts only a fraction of a percent of the AI investment boom, we’re talking hundreds of millions of dollars per year. And as we learn from the quantum story, money means people, and attention, and resources — and discoveries.
As I have written before, in my view consciousness is the most interesting scientific problem of our time, so I take this as happy news.
You may consider it a minor point amongst all the other wailing and gnashing of teeth about AI. But think of it this way. A century from now, and for many centuries beyond, school children will pay as much attention to the 21st century revolution in AI as they currently do to the invention of the light bulb, i.e., none. But the brightest minds among them may fizz and flourish at the startling explanation of the most incredible of all natural phenomena: human consciousness.
I can’t wait!
A few years ago I wrote a series of pieces about why consciousness is the biggest scientific problem of our time, and how I flirted with the topic in the late 1990s, and how disappointed I was to check back in 25 years later and discover that the progress during the intervening decades had been exactly zero. I learned that from Erik Hoel’s book on the subject, which was very nice, except that he foolishly believes in free will, but I kindly cleared that up. In his latest article he is more upbeat about progress in consciousness research, but I’ll stick with his earlier assessment, before he was fishing for funds.
My missives:
Erik’s latest rousing call to arms for consciousness research:
Somewhere in there he talks about consciousness research being “pre-paradigmatic science”. I agree. If you’re wondering what the hell that means, it comes from Kuhn’s half-brilliant-half-batshit Structure of Scientific Revolutions, which I conveniently also wrote a piece about.
Recent arrivals may be wondering where the light and lovely humorous fiction is. Apologies. This is more than just another puerile humour Substack. Trust me, it’s better this way. Now you can conclude your consumption of the latest publication and go out into the world and not only brag to your friends about how contentedly entertained you are — you can also pretend to be smarter.
For absolutely brand new readers: yes, most of the time I write comedic fiction about science and academia. That’s just the kind of intellectual smorgasbord we have here at The Fictional Aether. The full back catalog, of both fiction and non-fiction, is in the Table of Contents.
I was tempted to belligerently put every instance of AI in scare quotes, but there is a limit to how obnoxious even I can be. We may all object that it’s a long way from real intelligence, but I guess that’s what the word “artificial” is there for.
I feel no guilt about being prickly and protective about my ideas, inconsequential as they may be, so I want to make clear that this idea was floating in my head for some time before I read Hoel’s article. I came up with it all on my own! I was going to put the idea into a comment on his article, but then I thought, “I can string an entire Substack post out of this!” (Does that count as a second idea?)
Quantum, by Manjit Kumar. I am no expert on the history of quantum mechanics, so I have no idea if Kumar’s perspective is correct. I will hand-wring on this point in a later footnote.
Like any good physicist, I often encounter non-physicists who, when they hear my profession, protest at great length, “Oh, I always hated physics! It never made any sense to me!” I try to be polite to these people, but of course I can’t help but judge them as repugnant morons. What the hell is wrong with them? Now maybe I can empathise. I just have to think of my own reaction to classes on thermal and statistical physics. I glazed over instantly. None of it stuck. What was with all those physical properties with random letters, S and H and G and so on? What kind of sensible physical quantity is defined in terms of the logarithm of another one? The only thing I can remember is the famous opening of Goodstein’s States of matter: “Ludwig Boltzmann, who spent much of his life studying statistical mechanics, died in 1906, by his own hand. Paul Ehrenfest, carrying on the work, died similarly in 1933. Now it is our turn to study statistical mechanics.” I realise this is all just some ridiculous pathology of mine, much like the ridiculous pathology that forbids much of the human population from physics in general. Sorry. Needless to say, I was always disappointed that the story of quantum mechanics starts with blackbody radiation.
I have no idea where this version of the story sits within expert historical opinion, but I would be very surprised to hear that the economic context of the late 19th century played no role at all.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.