2 ms·
The problem is -- just like the food products the author uses as analogy -- there is no reasonable agreed-on line between "natural" and "unnatural", "homemade"
by cmrdporcupine 4y ago
The problem is -- just like the food products the author uses as analogy -- there is no reasonable agreed-on line between "natural" and "unnatural", "homemade" vs "factory", and "AI" vs "not-AI" because the processes at work are a gradient.
Is my spell or grammar check an AI? No, why not? What if I translate a word from another language while writing, using an automatic translation tool? What about the algorithms that boost (or don't boost) my post through various networks, and pick an audience for it? And the analytics tools I use? etc. etc. And all of this will just become murkier and murkier.
"AI" is itself a problematic term, since as many of us know many of the things so-marketed are little more than a bundle of heuristics or a complicated statistical model. And many of the "authoring" tools generating texts or images or code are really just copy-and-pasting-and-mutating things without a lot of magic in-between, but in clever ways that "trick" us into seeing it as original authorship. (Ok us humans do this too)
Perhaps the most reasonable dividing line is: is there a machine here that is emulating a human, pretending to be one? That is perhaps the thing that bothers me most about the recent wave with ChatGPT, Google's Meena etc -- the authors of these systems didn't have to create systems that presented with a human-like identity with pronouns and the illusion of selfhood. But they did, and now we have the Lemoine incident etc.
Reminds me very much of Dune, and its "Orange Catholic Bible"'s “Thou shalt not make a machine in the likeness of a human mind.” I think Herbert was onto something -- the question of whether something is truly human (or human created) or not might become the key important issue of our time and the ambiguity and confusion around that is going to become troublesome.