Posts by afamiglietti79@mastodon.social
(DIR) Post #ATdppi5FTdhdBqziCm by afamiglietti79@mastodon.social
2023-03-15T13:48:12Z
0 likes, 0 repeats
@TedUnderwood I like mine better. It's got that MS Paint Authenticity that the 2020s are gonna crave.
(DIR) Post #ATdq9yQGRGo3ekUEL2 by afamiglietti79@mastodon.social
2023-03-15T13:23:21Z
0 likes, 0 repeats
Please enjoy my Microsoft Paint rendering of the Most Important Curve in the World This Morning. Basically, LLM error reduction is going to follow an asymptotic curve. As developers make larger models and tweak feedback mechanisms, they will make rapid strides reducing confabulation, but that will inevitably level out at some point. The question is, at what point? If they level out a bit better than typical human, they eat the economy. A bit worse, and they are just Siri 2.0
(DIR) Post #ATgCgXRmLC9UrC32vY by afamiglietti79@mastodon.social
2023-03-16T17:06:16Z
0 likes, 0 repeats
@MattHodges @johnlray I mean, in theory, this sort of direct replication of training data ("memorization") is NOT supposed to happen. If it is happening more frequently than we think, that's a *major* scandal for the tech.
(DIR) Post #ATip7mXMWtH03DztwG by afamiglietti79@mastodon.social
2023-03-17T23:33:50Z
0 likes, 0 repeats
@TedUnderwood how do results compare to your previous model?
(DIR) Post #ATm7v7IgVBnFHmgbAm by afamiglietti79@mastodon.social
2023-03-19T13:48:34Z
0 likes, 0 repeats
@TedUnderwood @dh thanks for this, it's a really useful pointer towards method. It's also a helpful reminder that the default "chat" interface (where you just rely on *content* from GPTs training data) is mostly just a demo/party trick. The real power (and, accordingly, the real danger) is using the model's "understanding" of structure on more carefully selected problems and data.
(DIR) Post #AU1Nqqz6bDTvHDMvce by afamiglietti79@mastodon.social
2023-03-26T22:16:32Z
0 likes, 0 repeats
@jbj @TedUnderwood not mention that we'll need a *much* more forgiving political economy, given many attempts at adventurous knowledge creation will fail, since it's adventurous... And then there is the meaning problem. "I spent my whole life trying to imagine something no one else ever thought of, since machines do everything below that bar, and I never quite did" is going to be a real bear of a form of alienation...
(DIR) Post #AU5ieNAeUmuv0PV1LE by afamiglietti79@mastodon.social
2023-03-29T00:39:38Z
0 likes, 0 repeats
@TedUnderwood @benmschmidt you can't just unilaterally declare something DH... (checks notes)... It appears you can! Carry on.
(DIR) Post #AU7bDvSCc4t69Q4PgG by afamiglietti79@mastodon.social
2023-03-29T22:25:51Z
0 likes, 0 repeats
@TedUnderwood the sorts of left moralists who might have jumped in your @ replies have abandoned the platform?
(DIR) Post #AUPQBoWepV3mF4Z88e by afamiglietti79@mastodon.social
2023-04-07T12:45:07Z
0 likes, 0 repeats
@simon I agree but would go one further. It's not just a technical fix, we've got to educate people out of the mindset that it's possible to type a question in a box and get back "truth" in a neat package, easy to read, and less than 300 words...
(DIR) Post #AUPSiltTpsJ1Xn5GDI by afamiglietti79@mastodon.social
2023-04-07T12:51:10Z
0 likes, 0 repeats
@simon on the technical side though, I can't help but wonder, if we want accuracy and accountability, why have an LLM generate novel language at all? Why not do an embedding search of sources and just return excerpts (though maybe LLM summary could be useful here)
(DIR) Post #AUPU0WoVcZGZ5XvabI by afamiglietti79@mastodon.social
2023-04-07T13:26:48Z
0 likes, 0 repeats
@simon oh, I agree, there are uses for LLMs! Lots of them! I just wonder if the search engine like function people are eager to use them for (and encouraged to use them for) is in fact their best use
(DIR) Post #AURcz3Rz5opnzrZGcq by afamiglietti79@mastodon.social
2023-04-08T14:19:46Z
0 likes, 0 repeats
@TedUnderwood
(DIR) Post #AUTiceVY8UFKJEr9Mm by afamiglietti79@mastodon.social
2023-04-09T14:32:24Z
0 likes, 0 repeats
@TedUnderwood A reasonable take, though it leaves out the fact that building and participating in "meaningful learning engagements" is a LOT of work, for students and faculty alike, especially for those of us in teaching-focused roles at teaching-focused institutions
(DIR) Post #AUWr7OjcDDI5TTUF16 by afamiglietti79@mastodon.social
2023-04-11T00:21:59Z
0 likes, 0 repeats
@TedUnderwood a few days ago you posted about some sort of roll-your-own ChatGPT thing? I forgot to bookmark and I forgot what it was called, remind me?
(DIR) Post #AUWrGLC8k7Hj6N5qOO by afamiglietti79@mastodon.social
2023-04-11T02:53:21Z
0 likes, 0 repeats
@TedUnderwood that was it!
(DIR) Post #AUic53UKCUvwXDlXHc by afamiglietti79@mastodon.social
2023-04-16T18:59:39Z
0 likes, 0 repeats
@TedUnderwood @Jonathanglick It's weird that we're in a place where the statement "Let's not get ahead of ourselves, this will merely be as big a deal, socially, as the personal automobile was for the United States" is a moderate projection of a technology's impact, but I suppose we are...
(DIR) Post #AUieF1gDyfjRjPDJzc by afamiglietti79@mastodon.social
2023-04-16T19:20:54Z
0 likes, 0 repeats
@Jonathanglick @MattHodges @TedUnderwood I'll let Ted speak for himself here, but I think he intended that, and I agree with him. I've see the Discord servers where they congregate! Of course, the hippies weren't ethical angels, for example intentional communities were often racially segregated and gender essentialist, and AI art creators aren't either. But it won't be ALL Google's stomping boot on the human face either...
(DIR) Post #AUisekueL6W8cF7kx6 by afamiglietti79@mastodon.social
2023-04-16T22:05:30Z
0 likes, 0 repeats
@TedUnderwood @jose_eduardo @Jonathanglick @MattHodges hence why Ted's prediction is the high impact end of my error bars. The car was a Big Deal. Especially in the US! There is also pretty convincing argument to be made that the personal car and cities built around them was basically a terrible mistake. But that's part of what makes the analogy work so well. The danger is there too, and the possibilities of responding to the danger with policy...
(DIR) Post #AXUoc0mj1Ry534ZIuW by afamiglietti79@mastodon.social
2023-07-08T19:51:05Z
0 likes, 1 repeats
A flash-fiction response to the forum "Again Theory: A Forum on Language, Meaning, and Intent in the Time of Stochastic Parrots" organized by @mkirschenbaum (with work by @TedUnderwood and @elotroalex) at Critical Inquiry last week. It imagines what the central metaphor for machine language - poems washing up on a beach - might look like if it actually happened. Spoiler alert: individual interpretations of textual "meaning" are not a very important part of this story.https://afamiglietti.org/uncategorized/wordsworth-beach-a-flash-fiction-response-to-the-again-theory-forum/
(DIR) Post #AYatHCgXb2eO0OSwQi by afamiglietti79@mastodon.social
2023-08-10T16:30:47Z
1 likes, 0 repeats
@scalzi