A fried of mine asked “what is intelligence” on LinkedIn, linking it to being alive and wondering where “artificial intelligence” is on this spectrum.
For my own record, without further ado, my take:
Artificial intelligence (AI) is a misnomer, because it is not intelligence as we have it in humans. The behavior we observe in well-programmed AIs, including those based on LLMs, appears to exhibit intelligence, but it is not. We just ascribe the attribute “intelligence” to it. AIs still fail spectacularly where humans don’t.
An important part of human intelligence is that it is embodied in a human who exists in the world. Intelligence relies on a complex array of senses and information feeds that machines still don’t have. Naturally some may argue that robots are getting there, but I doubt they will get there any time soon. And again, what we will see in robots as sensing and act(uat)ing is likely to be a far cry from what humans can sense and act on.
Finally, intelligence relies on a gestation process in which humans mature and build out their sensing abilities, acting abilities, and reflection abilities i.e. intelligence. Neither pure-bit AIs nor embodied (in robots) AIs go through such a process. Humans prescribe by programming their training on how robots learn, and they always learn in a controlled environment to avoid that something goes wrong.
For reasons of the preservation of humanity I hope we will never let robots go through any such process like a maturing human, should we ever figure out how to do this.
In the meantime, none of these philosophical ruminations matter. We will use AIs where we can put them to good use, until the use is no good any longer, because they fail. There is nothing wrong with using tools, even if they appear to be intelligent.
Leave a Reply