Thanks for this. It's congruent with my own assessment. Human-equivalent artificial intelligence presumably is possible (although the motives for creating it are dubious), but I think it's far more difficult than many people still seem to believe, even after all the hype cycles the field has gone through since the 60s. Existing "AI" remains shallow and brittle. (For what it's worth, I've been observing the field since the 80s. I've designed, built, and used machine-learning systems for commercial and scientific purposes.)
If he were still around, Jacques Derrida could have a lot of fun with this kind of thing. A rude, crude summary of some "deconstructionist" thinking is, "Meaning is imported to texts by the interpreter's will to power." In this case, it seems, meaning is imported to texts by the interpreter's will to attribute sentience to computers.
Thanks for this. It's congruent with my own assessment. Human-equivalent artificial intelligence presumably is possible (although the motives for creating it are dubious), but I think it's far more difficult than many people still seem to believe, even after all the hype cycles the field has gone through since the 60s. Existing "AI" remains shallow and brittle. (For what it's worth, I've been observing the field since the 80s. I've designed, built, and used machine-learning systems for commercial and scientific purposes.)
If he were still around, Jacques Derrida could have a lot of fun with this kind of thing. A rude, crude summary of some "deconstructionist" thinking is, "Meaning is imported to texts by the interpreter's will to power." In this case, it seems, meaning is imported to texts by the interpreter's will to attribute sentience to computers.
this remains, if not my favorite substack, then definitely the most-disquieting one i subscribe to