<aside> <img src="/icons/help-alternate_gray.svg" alt="/icons/help-alternate_gray.svg" width="40px" />
A rough draft of thoughts about the problems with the definition of Artificial Intelligence through comparing it to human intelligence.
Written by Dominik Lukeš. Please send comments to [email protected] or https://x.com/techczech.
</aside>
It is very common to see this sort of thing as definition of Artificial Intelligence:
Artificial intelligence is a computer system designed to perform actions for which human intelligence is typically required.
Why is this wrong? For the very simple reason is that it describes almost anything a computer can do. In fact, it even describes a calculator. Multiplying numbers is certainly something for which human intelligence is required. And it pretty much also describes a dictionary. Because we need intelligence to remember words.
In fact, defining Artificial Intelligence by starting from a definition of intelligence is the least helpful way of going about it. It would be much more accurate to say that AI is the sort of thing Artificial Intelligence researchers are trying to make computers do. And this will explain why what AI means has changed through its roughly 70-year history.
But we can also start looking at understanding Artificial Intelligence by thinking about what is left over when we've designed a machine to replace some aspect of what humans do.
This reflection was inspired by a quote shared on X from a Bloomberg Television interview with Geoffrey Hinton:
In the industrial revolution, we made human strength irrelevant. Now we're making human intelligence irrelevant. And that's very scary. (AI's 'Existential Threat' to Humans - YouTube)
Some people took issue with it on humanist grounds:
Tell that to all people working physically demanding jobs. Such an ivory tower comment.
But he's actually very much wrong factually. The contribution of industrial revolution was not make the need for human strength go away. What it did was make the contribution of any individual human to what was produced much less but at the same time make that contribution more valuable because of what the human in charge of a machine could produce.
Ever since the invention of agriculture, human strength was mostly irrelevant. That's why humans domesticated animals, invented things like the plow, the lever, the wheel, etc. Because human strength was never enough. And the history of technology was always characterized by making up for the humans' deficiencies in strength, speed and endurance.
And it was also always accompanied by the worry that technology would be too much, that it would make humans too soft, take away their jobs, deprive them of a sense of purpose, and ultimately, completely replace them (and possibly suffer them as pets).
So, Hinton was wrong to say that it was the industrial revolution that made human strength irrelevant but his wrongness fell off the tongue so easily because in the last two hundred years, we have seen humans replaced by technology. Over and over and over. There are now several orders of magnitude more of humans than there were in the 1800s and even the poorest of them are better off. Yet, many fewer (as a percentage and often in absolute terms) work in professions that defined human pursuit of survival for centuries. Some are gone completely, others greatly reduced in number and importance and others survive merely as hobbies or artistic pursuits.
Yet, there is still plenty of work around where no machines can tread. And there is no area of human endeavor where humans play no role at all. We might like to think that it is intelligence that is so vital, but I'd like to argue that it is the triumvirate of dexterity, judgment and persistent identity that makes humans (so far) indispensable. Let's take them in turn.
While human strength is relatively feeble compared to that of even a simple lever, dexterity is a different matter. While we are often clumsy and fumble with the world around us, we are the most dexterous of animals and far more dexterous than any machine yet built.