Even if I am not all that interested in ChatGPT or AI in general, one cannot go around today without hearing about it. Can AI eventually replace humans? I think the answer is a foregone conclusion. Yes, of course. So far, the text sputtered out from ChatGPT is indistinguishable from a mediocre human writer: long-winded, dull, unreliable, a mixture of truth and rumors and occasional lies.
The heart of the matter is not whether AI is in fact human-like, possessing the goodies that we have assigned to ourselves: consciousness, intelligence, self-awareness, etc. The real question is: Can we tell the difference? That's the essence of the Turing Test, and the answer is clearly, NO. A human cannot look at blocks of text on the screen and know whether it has been written by another human or a machine.
So the uncomfortable question is: Are human no more than a machine? More than a possible future in which humans are obliterated by machines, the fear and discomfort is in the here and now: We are not special and maybe not as smart as we think.
Sure, we can pat ourselves on the back for the fact that these human imitations were "created" or "invented" by humans, followed by the realization that we are not so complex or sublime or mythical that we cannot be easily imitated with some codes and billions and trillions of human-generated words as training material. In other words, real humans are pretty boring and predictable. We're also a bunch of narcissists who molded some clay creatures in our own image, and now in these images we see the plainness of ourselves. Oops.
1 comment:
Both of your postings about AI are really good and thoughtful, thanks. You're right that AI can never be human however good it becomes at pretending to be so--and personally I am not impressed with some of what chatGPT creates at least now. A friend of mine asked it to create a poem in the style of e.e. cummings--what it produced was a joke, nothing like the cleverness and mischievousness of cummings. (On the other hand I asked it to write a two paragraph lede in journalistic style about the possible uses and effects of AI in psychiatry--and it was pretty good). But you are especially right in pointing out that AI should be humbling us, regardless. "So the uncomfortable question is: Are human no more than a machine? More than a possible future in which humans are obliterated by machines, the fear and discomfort is in the here and now: We are not special and maybe not as smart as we think." Just so. But i will only add that just as AI is constantly learning and improving, so could we be--but its a choice to do so, to commit to that. Most of us dont want to try and all of us live in systems that discourage actually living up to our humanity, or actively make it difficult or impossible.
Post a Comment