The World’s Smartest A.I. Is Still Dumber Than a Baby

An interview with the French neuroscientist Stanislas Dehaene, author of ‘How We Learn: Why Brains Learn Better Than Any Machine… For Now’

Image: Westend61/Getty Images

AAlgorithms have thrown the gauntlet down. They’re challenging our distinctive status as the most advanced learning species on the planet. In the past several years, machines have “learned” to instantaneously transcribe a foreign language and detect typos in our Google Docs; they’ve predicted the superfecta of the Kentucky Derby, provided well-wrought medical advice, composed classical music albums, and humbled us at chess. And yet, according to the French neuroscientist Stanislas Dehaene, the most sophisticated artificial intelligence technologies are still far less smart than the learning capabilities contained in even an infant’s brain.

While the iPhone’s voice-controlled assistant Siri can recognize and “learn” a word, a feat that takes a multitude of training attempts, a slew of big data, and high-power servers, a young child can learn one in a repetition or two. In his new book, How We Learn: Why Brains Learn Better Than Any Machine… For Now, Dehaene grapples with how humans learn, and ways engineers are attempting to use A.I. — and still falling short — to mimic our learning abilities.

According to Dehaene, one of many reasons we’re still at a higher level over machines is that our minds act as superior statisticians. Specifically, he contends our brains developed “algorithms” via evolution — ones that constantly attend to uncertainties and probabilities — and at far greater capabilities than A.I. currently can.

“Learning is grasping a fragment of reality, catching it, and bringing it inside or brains,” he says. And, he maintains, this unique process is the highpoint of humanity.

Dehaene is a professor of experimental cognitive psychology at the Collège de France, as well as the director of the Cognitive Neuroimaging Unit in Saclay, France. OneZero caught up with him to talk about brains both human and otherwise, the impact Google has on our capacity to learn, why sleeping is so important for solving problems, and what a blind mathematician taught him about learning.

This interview has been edited and condensed for clarity.

OneZero: You argue that even the most advanced computer architectures are less effective than our brains, even infant brains. How so?

Stanislas Dehaene: I don’t want to downplay the successes of A.I. In fact, I cite in the book some nice work showing that there is a correspondence between the architectures of A.I.’s neural networks [that is, algorithms modeled after the human brain] that are designed to recognize patterns like the first stages of our vision. But that only works for about the first quarter of a second, which is essentially the unconscious processing pathway in our brain. So as soon as we are beyond this first quarter of a second, the human brain seems to be able to do better than the algorithm.

In humans, I think we have yet another better level of processing, which is symbolic processing. We are able to extract information, and this is where the power of human learning really becomes extraordinary. We extract from the world, not just implicit information like neural networks do, but actually explicit information that we can share with others in the form of symbols that we haven’t really formulated in language. We do it with scientific theory. So one of the main messages of my book is that even a young child is already formulating a sort of scientific theory of the world.

What do you think about our ability now to retrieve information so quickly, say by Googling something? How has that affected our ability to learn?

I think humanity has always relied on additional devices in order to improve its learning. It’s rather remarkable the whole invention of the school and the education system is a cultural device to enhance the learning in our brains, and also to fill some gaps in our capacities. So I think Google is just another cultural invention that we have at our disposal. It does not prevent learning. It enhances learning because you have so much more information that’s available. Of course, in order to understand what Google is telling you, you still need the fundamentals. So more than ever we need to revert to reading and vocabulary. We need arithmetic. A sense of numbers. I don’t believe much in the so-called 21st-century skills. I still think that we need to focus our brains and the brains of our children on these fundamentals.

The “for now” part of your book’s title seems to suggest algorithms one day will learn better than humans.

I think that there is no limiting principle for computers. But I think it’s the same for human intelligence. When you look at the human brain, I’m completely awed by the brain of a young child. It’s a supercomputer. And look at the amount of energy that the brain consumes — 20 watts. That’s not even enough to light up an old-fashioned light bulb. But it’s enough to power this incredible supercomputer. It’s at least a million times less than any of the machines that are comparable in power in some respect in computational power. So I think it’s going to take a long time before we can imitate these sorts of properties of compactness and efficiency.

What do you make of the amount of time we spend today looking at computer screens — from smartphones to laptops? Do you think that’s affected our ability to learn, and do you think unplugging from those things assists with learning more efficiently?

I don’t think we can speak of screens as one entity. It all depends on the software and the abilities you are using while interacting with it. You can perfectly read on a screen or you can play your video game and there are some extremely powerful educational video games that require unbelievable attention.

The one danger of our use of screens at the moment is that there are a lot of distractions and our brains are just as susceptible as ever to the power of distraction.

We cannot multitask. We can only process consciously one piece of information at a time. The problem is that there is all this extremely sophisticated software which is actively trying to distract us and make us addicted to one more distraction. So this is powerful software. It’s designed to grab your attention and not release it. So I think if we learn to use software wisely, and especially in terms of the time that we spend on it, then it can be an extremely powerful assistant. There is beautiful research showing that people who play video games can develop better attention and decision-making. So this can be very useful.

What is your stance on so-called “brain hacking,” for instance, people without ADHD taking Adderall to help them focus or the trend of micro-dosing LSD to be more creative?

I think it’s dangerous, especially using things that are outside of a clinically proven protocol. Our brain is the most precious thing. What we know is it is a very stable device, but it needs good food, oxygenation, and sleep. I spend a lot of time speaking about sleep because I think we understand much better why it is important for the brain. But so I would rather manipulate my sleep than my brain with drugs.

Why is sleep so important for learning?

The brain actually repeats information during sleep. So even though you have the impression of being absent and just resting, the brain is not. The brain is very active, and it replaces information and the neural activation patterns that happened in the previous days. Especially the important information and tasks you’ve been doing, the bits that were labeled with emotion.

So what happens is that you have these slow patterns during the day, and during the night you replay them in your brain, except about 15 or 20 times faster, which allows your brain to rehearse a lot more than during the day, and as result helps you solve tasks you were stuck on. And I think this is one of the very important tricks that evolution has found and implemented in our brain, the ability to rehearse unconsciously at a faster pace than we can do normally. In a sense, it is rehearsal by imagination, and it exists in all sorts of species. It’s a very important learning mechanism.

Finally, how can perceived handicaps help us better understand our learning processes?

In the book, I mention a number of cases of patients with deficits or handicaps, and yet they still manage to learn. The fact, for instance, that you can be blind and still be a mathematician. I think this is extremely important because it means that there is a resilience of the human brain. Learning allows us to go beyond the limits of one particular sense modality.

That’s a message of hope — that learning shouldn’t be underestimated. There are many parents or teachers who think that because there is a handicap, in a given child, we should know where our expectations are. I don’t think that’s the case and when I met this extraordinarily gifted mathematician who was blind from very early on, this really changed my ideas about how much we can achieve in the human brain.

A writer. Not based in Brooklyn. Recent bylines with Vox, Vanity Fair, Harvard Magazine, MIT’s Undark, VICE and Playboy.

Sign up for Pattern Matching

By OneZero

A newsletter that puts the week's most compelling tech stories in context, by OneZero senior writer Will Oremus. Take a look.

By signing up, you will create a Medium account if you don’t already have one. Review our Privacy Policy for more information about our privacy practices.

Check your inbox
Medium sent you an email at to complete your subscription.

The undercurrents of the future. A publication from Medium about technology and people.