On the limits of human intelligence

A dog lives in the same universe as Albert Einstein.

It experiences gravity, time, motion and space. It knows when something falls, when the sun rises, and when dinner is late. It navigates the physical world remarkably well.

But you cannot teach a dog calculus.

No amount of patience, repetition or encouragement will get you there. The problem isn’t a lack of information. The dog’s brain simply lacks the cognitive architecture necessary to understand mathematics.

Now consider ourselves.

We have spent thousands of years trying to understand the universe. We have developed mathematics, physics, philosophy and increasingly sophisticated instruments to extend our senses. We have discovered atoms, galaxies, black holes and the strange behaviour of particles at the quantum level.

And we have made a rather extraordinary assumption along the way.

That the universe is ultimately comprehensible to the human mind.

We might not understand everything today, but we assume that, given enough time, someone eventually will.

Why?

Evolution did not produce the human brain to understand the universe. It produced a brain capable of keeping a rather vulnerable primate alive long enough to reproduce. Our ability to develop mathematics, write symphonies and contemplate the origins of existence is an astonishing consequence of that evolutionary process.

But there is no particular reason to believe that evolution equipped us with unlimited intellectual capacity.

We may simply be cleverer dogs.

Now imagine an artificial intelligence, perhaps a century from now, that finally discovers how the physical universe works.

It reconciles quantum mechanics and gravity. It explains the nature of time, the origins of the universe and why the fundamental constants of physics have the values they do.

It produces predictions. We test them. They prove correct, repeatedly.

We ask the machine to explain its discovery.

And it tries.

It simplifies the mathematics. It creates analogies, visualisations and elaborate teaching methods. It patiently works through the explanation, again and again.

But we cannot understand it.

Not because the machine is deliberately obscure. Not because it is withholding information. But because the concepts required to comprehend its discovery lie beyond the capabilities of the human brain.

Just as calculus lies beyond the capabilities of a dog.

There is, of course, another possibility. AI might discover entirely new ways of teaching us, extending human understanding into territories we cannot presently imagine. We don’t know whether our cognitive limitations represent an absolute ceiling.

But suppose they do.

We would find ourselves in an extraordinary position.

We could know that the machine was right without understanding why it was right.

Humanity would possess an explanation of the universe that humanity itself could not comprehend.

For centuries, we have imagined that the ultimate achievement of science would be a complete human understanding of reality.

Perhaps we have misunderstood the destination.

Perhaps the ultimate achievement of human intelligence will be the creation of an intelligence capable of going where we cannot follow.

There is something wonderfully absurd about that.

We began by believing the universe was created for us. We gradually discovered that we occupied no particularly privileged position within it. Our planet was not the centre of the solar system. Our sun was not the centre of the galaxy. Our galaxy was merely one among billions.

And now we may be approaching the final indignity.

The discovery that our intelligence is not particularly privileged either.

The ultimate triumph of human intelligence might be discovering the limits of human intelligence.

Somewhere, a dog is watching its owner struggle with a mathematics textbook.

It has no idea what all the fuss is about.

Perhaps, one day, neither will we.