News for hackers. Stuff that matters.
Re:Have I reached the Douglas Adams Inflection point? by jimbokun on Thursday September 15, 2022 @11:29AM (#32853560)
Which is strange because that's exactly what very large language models like GPT-3 are good at. Keeping the context and generating plausible next sentences, paragraphs, etc.

Wonder what's blocking that from being applied to systems like Alexa.

Re:Have I reached the Douglas Adams Inflection point? by groby_b on Thursday September 15, 2022 @04:04PM (#32857754)
Unpredictability.

You (Amazon) don't want your precious little bot say random things that land you in the press. And oh, GPT-3 will.

Point two is that models that can be goaded to a specific place. So you'll need to account for that too, or... oh look, press.

Point 3 is that they're trained on the Internet at large, and you don't want that in your house :)

Re:Have I reached the Douglas Adams Inflection point? by randcraw on Thursday September 15, 2022 @04:35PM (#32858103)
And if it chooses the wrong prior topic (among several possibilities) the smartbot could behave maddeningly obtusely. I suspect even the best of these today will have little idea what part of the conversation is more relevant to the human's focus than any other part. It's just as likely to turn its attention to an irrelevant 'token' that came before as a relevant one and say something obvious or fatuous.

It's a lot easier to converse gracefully than to add useful info to a conversation.

This is essentially what has made past intelligent assistants so dysfunctional (like Microsoft's Clippy). They couldn't model cognitive concepts like 'relevant', 'interesting', or 'necessary' relative to the active topic.