This article is more powerful than the human brain.
A news report suggests Meta plan to develop an AI that is more powerful than the human brain. But anything can be more powerful than anything if you don't care about the specifics.
I was recently part of a roundtable discussion with journalists, union members, high-up politicians, and more, to explore the increasingly-precarious state of modern media, with regards to the work of freelancers and how they’re paid. Or not, as is the case bleakly often.
Predictably, and unavoidably, the discussion quickly turned to AI, and its increasing omnipresence in the media world, which often involves it ‘taking’ the work of others, often the barely-getting-by freelancers just mentioned, for zero compensation, and regurgitating different variations of it, that are increasingly used as copy for news platforms etc., thus denying freelancers and contributors the opportunity to earn a living twice over.
It’s not great. Hopefully something will be done about it. And soon.
But as I was on the train to the meeting, I saw a post which encapsulates the whole issue perfectly. It was about this story for The New York Times. And here is how it was presented online (on BlueSky, in this instance).
So, let’s break this down. Not even line by line, but word by word.
“Breaking News” means that this story is deemed to be so important that it warrants being published and promoted right away, ahead of the pre-planned output. Big media platforms tend to schedule what they publish quite carefully, so one must assume THE NEW YORK TIMES, one of if not the most prestigious newspaper there is, finds this story to be of crucial important. Is it? Let’s keep going and find out.
“…:Meta…”, the Zuckerberg-owned tech company. I’ve made it clear how I feel about them, more than once.
“…is said”, …so this isn’t a definite thing? Not a press release or official statement? It’s just a rumour?1
“…to be preparing…” So, they haven’t actually done the thing that it’s rumoured they’re doing? Not yet, at any rate.
“…to unveil…” OK, so this isn’t a rumour about a tech company making preparations to do something. It’s a rumour about a tech company making preparations to show us something. Or just announce it. I’d always assumed that to be deemed ‘news’, something had to actually happen. But according to this, that’s wrong, because all this seems many steps removed from an actual event taking place.
“…an AI lab dedicated to a hypothetical system…” Good lord. So on top of everything else, the thing at the centre of all this very-tenuous news doesn’t actually exist, and may never do?
“…that exceeds the powers of the human brain”. …we’ll get back to this in a moment.
So, in summary, what we have here is a rumour about possible preparation, to announce an effort to develop a system that doesn’t and maybe cannot exist, with capabilities that exceed some arbitrary benchmark.
Exactly how this counts as news at all, let alone ‘breaking’ news, is hard to explain. Unless they mean “this story is so insubstantial it literally breaks the concept of news”? Points for honesty if so, I guess.
But this is what we’re up against. One of the most powerful tech companies on Earth might do something with AI, and one of the most influential media platforms drops everything to tell the world about it. Presumably resulting in a leap in stock prices for the former and a healthy blast of traffic and ad revenue or whatever for the latter. All without anyone having actually done anything of substance.
What exactly is a lowly freelancer meant to do in the face of all that?
Well, we persevere! Because what else is there.
And this freelancer is here to take issue with one aspect in particular; the notion that this purely theoretical AI would be ‘more powerful than the human brain’.
Because, unnerving as it may sound, it’s pure propaganda and bluster. It’s one of those phrases that sounds intimidatingly impressive when you first hear it, but upon looking closer, you realise it doesn’t mean anything at all. And also, not anywhere near as significant as it seems. It’s like when people say something is a ‘quantum leap’ forward, relying on the fact that this is arguably the smallest possible movement in the known universe.
Basically, to suggest that you’ll develop an AI ‘more powerful’ than the human brain is simultaneously impossible to clarify, and functionally meaningless.
And, from certain perspectives, it’s not even an achievement!
How powerful is the brain, actually?
Point is, how ‘powerful’ is the human brain? Because it’s not an especially specific term.
For instance, power is often described as the ‘rate at which energy is transferred or consumed’, measured in watts. And the human brain tends to do all it needs to do on 20 watts of power. The typical AI system (or ‘Large Language Model’, LLM, to be more precise) requires orders of magnitude more power than that, just to function.
So, technically, just by existing at all, any working AI programme is already more ‘powerful’ than the human brain.
Presumably, this isn’t what Meta or The New York Times actually means. But it does give them an ‘out’, if anyone wants to question this ‘more powerful than the brain’ claim.
One would assume they actually mean this theoretical AI could ‘do more stuff’, or be ‘smarter’ than the standard human brain. However, this is also objectively meaningless.
Like I’ve said before, quantifying how intelligent or ‘smart’ someone, and thereby their brain, is, that’s already a very slippery and uncertain process. But in terms of raw numbers, it could be said that the human brain isn’t that powerful at all.
When it comes to handling and manipulating raw data, it’s believed that our working memory, the brain process which does just that, can hold around 4 items at once. Meanwhile, I’m no tech expert, but if an AI is trained with, and can respond by drawing on, all the available works on the web (even those they don’t have permission to use) then it logically must have a capacity far in excess of ‘four bits of information’.
By that logic, any computational device is more powerful than the human brain. Indeed, this article is more powerful than the human brain (hence the title). It can clearly hold thousands of words in a specific sequence, for an indefinite period. And I can refresh it as much as I like, it’ll recall them all perfectly. Find me a brain that can do that from the get go, without spending many hours learning and reciting the text at length. Evidence shows that human memory is mostly about generalities or the ‘gist’ of events, unless it’s something incredibly stimulating (i.e. traumatising) or extensively rehearsed.
Imagery, too. If you asked the typical human brain to recreate an image of, say, a cheeseburger, it could remember or figure out what one looks like, but it would struggle to recreate that image with 100% accuracy. It’d be some vague, muddy representation that would be difficult to recreate in a tangible sense.
Meanwhile, look!
There, a perfect representation of a cheeseburger, right down to the reflection of the light on the shiny processed cheese slice. See, this article is indeed much more powerful than a human brain, let alone an advanced AI.
You might be saying “But that’s not what they mean!” And you’re probably right. But that’s the problem; we can’t say with any certainty what they do mean, because they haven’t clarified anything. It’s just a bold, impressive sounding claim which can mean whatever you want it to, if you’re so minded.
Do they mean they’ll create an AI that’s more intelligent than a human brain? If so, in what way? If it can manipulate more data at once, then that’s easy. Most modern computers do that.
If they mean it’ll operate faster than the human brain, well, yeah. It would be weird if it didn’t. The maximum speed of transmission of information in brain cells is about half the speed of sound. In modern electrical computing devices, it’s a good chunk of lightspeed. The latter is slightly faster than the other, shall we say. That’s part of the reason why pocket calculators became so popular in the first place. We could do the mental arithmetic required if we put our minds to it, but it’s so much quicker to let the gadget put the legwork in.
If they mean they want to make an AI more efficient than the human brain, then… great! That 20 watts of power I mentioned just now? The human brain uses that to perform trillions of calculations. If Meta want to create an AI that can do everything an AI is meant to do with all the power requirements of an old-school light bulb, then more power to them (ironically).
If they mean they want to create an AI that can somehow combine all the myriad data processing functions they perform into a unified consciousness that is self-aware and capable of self-adjustment, learning, and deduction, without external modifications or input… cool, I guess? But they really should let us know how they did that if they manage it, because this is a function of the human brain that we’re still a long way from figuring out.
The point is, there are so many things even a basic technological or software construct can do that the human brain cannot. And there are just as many things that the human brain can do that even the most advanced technology cannot begin to match. Saying one is more ‘powerful’ than the other is ultimately meaningless.
It’s like saying Mario Kart is a better game than Chess, because it has better graphics. In some ways, Mario Kart is a better game than Chess. In others, it isn’t. But chess doesn’t even have graphics! So this comparison makes no sense.
I get that it’s hyperbole and hype. But hype based on, that relies on, ignorance, is not hype I can ever condone.
Why not make your own brain more ‘powerful’ by buying one of, or several, of my books. They are well good.
I could, I suppose, read the article itself to find out, but I don’t have an NYT subscription. And once you’ve read to the end of this article, you’ll hopefully understand why I don’t see that changing any time soon.
This is a refreshing read, I'm getting a bit tired of the AI panic. I know we should be wary and as with all new technology it needs some level of management. But the sensationalism is getting boring, and I have faith that there are enough people with sense to navigate this new playing field.