Jordan-117

16 ms·

271.
by Jordan-117 2y ago
I think the default prompt attempts something like that -- one host does tend to take a more questioning role, and I've even had a few pods introduce one of them as a guest expert. But it seems like it sometimes loses track of who&#x27
272.
by Jordan-117 2y ago
I think there was a black-and-white photocopy from the early 90s, but the full-color digital one (with an iPad app!) is circa 2013.
273.
by Jordan-117 2y ago
This is a cool resource, but I'm side-eyeing them characterizing it as some new thing when this scan was uploaded more than ten years ago (as their own link to the college's archived blog post on it shows).
274.
by Jordan-117 2y ago
It's the RLHF training to make them squeaky clean and preternaturally helpful. Pretty sure without those filters and with the right fine-tuning you could have it reliably clone any writing style.
275.
by Jordan-117 2y ago
I've been using Ember with standard voice mode because it nostalgically reminds me of the style of narration from certain old 90s edu-games, but it doesn't register as "black" to me. Ditto what you term the "Sal Kha
276.
by Jordan-117 2y ago
Because if OpenAI's technology is as powerful and world-shaking as Altman says, it would be nice for it to be under the control of a nominally pro-social and humanist non-profit organization with safety and equity as paramount goals ra
277.
by Jordan-117 2y ago
It is a creativity test, but doesn't how you get there matter? From what I understand, aphantasiacs think in words, verbalized concepts, remembered facts, relationships, etc. So I'd expect that to make a visual analogy they'd
278.
by Jordan-117 2y ago
I feel similarly, though less about the lack of objective testing and more about the inability to even be sure we're talking about the same thing. Like I can visualize stuff fine, but not literally see it like some Hololens overlay --
279.
by Jordan-117 2y ago
You might enjoy Vivarium if you haven't seen it already.
280.
by Jordan-117 2y ago
Kane's spiritual sequel "The Oldest View" is kind of like this.
281.
by Jordan-117 2y ago
That's how it used to do it, but my understanding is that this new model processes audio directly. If it were a music generator, the original would have generated sheet music to send to a synthesizer (text to speech), while now it can
282.
by Jordan-117 2y ago
Seems like it was picking up on the audience reaction and stopping to listen. To me the more troubling thing was the apparent hallucination (saying it sees the equation before he wrote it, commenting on an outfit when the camera was down, d
283.
by Jordan-117 2y ago
Deep Philosophical Video: https://www.youtube.com/watch?v=ybZ9PVcf6X8
284.
by Jordan-117 2y ago
Sorry, meant the difference between this HN post and the article headline.
285.
by Jordan-117 2y ago
Odd that this says "X" while the article says "Twitter." Did the poster change the headline, or did USA Today?
286.
by Jordan-117 3y ago
It's not just a problem for training, but the end user, too. There are so many times that I've tried to ask a question or request a summary for a long article only to be told it can't read it itself, so you have to copy-paste
287.
by Jordan-117 3y ago
https://freecow.ytmnd.com/ https://takemeoutflanders.ytmnd.com/ https://finalstaringcontest.ytmnd.com/ https://vaderloveschristmas.ytmnd.com/ https://sixtymins.y
288.
by Jordan-117 3y ago
I used to scoff at the idea of paying for YouTube, but I'd absolutely consider it if they had a cheaper plan that excluded YouTube Music.
289.
by Jordan-117 3y ago
Imagine something like GAN Theft Auto V powered by this technology: https://www.youtube.com/watch?v=udPY5rQVoW0
290.
by Jordan-117 3y ago
I don't understand why they don't just include a dumb filter on output that blocks any repetition of the system prompt. Presumably they don't want these rules publicly known, yet they leak regularly.
291.
by Jordan-117 3y ago
Yahoo Answers dead, Quora dying, StackExchange struggling with AI spam and mod rebellions. Meanwhile [Ask MetaFilter]( https://ask.metafilter.com/ ) just turned 20 and keeps on trucking. Smaller scale obviously ( human -scale
292.
by Jordan-117 3y ago
Yes, it was the official Personal Blocklist extension: https://searchengineland.com/googles-personal-blocklist-dele...
293.
by Jordan-117 3y ago
Google used to have a feature where you could filter out whole domains for any reason. I wonder why they don't bring that back and then use that as a (non-determinative) signal for deciding which domains to ban themselves. They index a
294.
by Jordan-117 3y ago
The Guardian's recent follow-up suggests it's not something they track at all: https://www.theguardian.com/technology/2023/dec/30/reddit-mo... > While traffic has not changed substantially,
295.
by Jordan-117 3y ago
Monitoring social media for sentiment and automatically replying seem useful for sure, but the examples given of AI bots writing ostensibly personal recommendations under the guise of regular accounts without disclosing that they're pa