Here’s a longish read (NYT, but if you’re not already behind the paywall you all know how to burrow beneath it) on the advances that are being made in language manipulation by machine. You know me: I’ve long maintained that AI will creep up on us, and that artificial sentience (internally quite alien to our own consciousness but, as here, capable of impressive feats of mimicry) will arrive not with a thunderclap but incrementally, going variously unrecognized or denied long after the fact.
Anyway, this’ll be well worth your fifteen minutes. We’ve come a long, long way from ELIZA:
Anyway, this’ll be well worth your fifteen minutes. We’ve come a long, long way from ELIZA:
However the training problem is addressed in the years to come, GPT-3 and its peers have made one astonishing thing clear: The machines have acquired language. The ability to express ourselves in complex prose has always been one of our defining magic tricks as a species. Until now, if you wanted a system to generate complex, syntactically coherent thoughts, you needed humans to do the work. Now, for the first time, the computers can do it, too. Even if you accept the Gary Marcus critique — that the large language models simply present the illusion of intelligence, a statistical sleight of hand — there’s something undeniably momentous in the fact that we have finally met another magician.cordially,