Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI's GPT-3 may be the biggest thing since Bitcoin

maraoz.com

Post reply on HN
Print view View original

391–400 of 554 posts

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#391

Post by amelius »

geofft wrote: post #238
6nf wrote: post #233

Earlier quoted context omitted.

How many attempts did it take or did you just choose the first one? I have to admit, this is passing my turing test...

Really? I got about halfway through and realized that the comment had no point . If you tried to summarize what it was arguing, beyond the first sentence, I don't think you could make a coherent summary. Maybe the real lesson is we don't expect human-written comments on discussion fora to be particularly coherent....

I don't think these gpt3 comments will get many upvotes on HN anyway. I downvoted the first one for being incoherent, but then realized it was meant as an example so I undowned it.

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#392

Post by DonHopkins »

DonHopkins wrote: post #387

Reminds me of the time I fed RMS's infamous Natalism Flame into the Eliza-like Doctor "AI" in Emacs: https://www.emacswiki.org/emacs/EmacsDoctor >Emacs Doctor >When life, the universe and everything is all getting too much, Emacs has doctor.el to help. To consult the doctor type >M-x doctor >The doctor is a Rogerian psychotherapist who will help you with your problems. It would spoil the fun and hurt your recovery to…

Of all the NSFW responses, Lile Elam's shows the most dang-like superhuman patience, moderation, and kindness:

>Please send your "fucks" via personal mail and refrain from using Kabuki-west for such messages. -Lile Elam

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#393

Post by amelius »

pjc50 wrote: post #370
thewarrior wrote: post #133

Earlier quoted context omitted.

I've started working on a version of GPT-2 which generates English text. The purpose of this is to improve its ability to predict the next character in a text, by having it learn 'grammatical rules' for English. It already works well for predicting the next character when it has seen only a small amount of text, but becomes less accurate as the amount of training text increases. I have managed to improve this by havi…

> Plot twist: This comment was generated by GPT-3 prompted with some of the comments in this thread. This kills the forum. Seriously, once this is weaponised, discussion of politics on the internet with strangers becomes completely pointless instead of just mostly pointless. You could potentially convince a human; you can't convince a neural net that isn't in learning mode.

Perhaps it will destroy anonymity. Because the only way to be sure a human wrote something is if you somehow know who the comment came from.

We might end up with reputation based conversations.

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#395

Post by solidasparagus »

visarga wrote: post #228
solidasparagus wrote: post #34

Earlier quoted context omitted.

> What are humans except "text generation" machines? The ability to grow and choose our own direction: to choose what our goals are, curiosity, self-awareness, desire. To imply that GPT-3 is anything close to strong AI is kind of ridiculous.

GPT-3 has flexible goals too. It can learn a new task and do it in the same step. What GPT3 doesn't have is a body.

No, you can teach it a new task or to do multiple tasks, but it will never be able to independently identify what a new task might be and learn to do it. An important distinction when talking about AGI.

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#397

Post by DonHopkins »

meredydd wrote: post #28

I am deeply enjoying this comment thread - it's a bit of a Barium Meal [0] for determining how many people read (a) the headline, (b) the first paragraph, or (c) the whole thing before jumping straight into the compose box. Having read to the bottom, the quality of text generation there absolutely blew me away. GPT-2 texts have a somewhat disconnected quality - "it only makes sense if you're not really paying attenti…

It reminds me of the incoherent demented ramblings we've all been been hearing (but hopefully not following as medical advice) for the past several years.

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#398

Post by vladTheInhaler »

Alex3917 wrote: post #114
dannyw wrote: post #88

Earlier quoted context omitted.

If that’s the only thing that separates this from human writing; I’m sure it can be influenced easily.

> I'm sure it can be influenced easily. Maybe. Right now this reads like a glorified shopping list. It's coherent, but actually sounding human also requires a theory of mind. E.g. I explain here why it's possible for written statements to be objectively insightful, informative, interesting, or funny, but objectively in a way that's relational to other information or beliefs. The implication being that statements are…

I think the issue is that text doesn't exist in a vacuum, but the corpus that the model is learning from does. A piece of human writing exists for a particular reason - to persuade, to inform, to ask a question, etc - and its value is judged on its ability to perform that task. But that's not a quality that is evident from the text itself, only from looking at the world outside the text. This suggests to me some limits on this kind of passive self-supervised approach. Perhaps it could be improved by augmenting the text with other forms of data? For instance predicting video from text and vice versa. But I think that to learn a true "theory of mind", it needs to use text like an agent - to influence its environment, not merely predict it.

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#399

Post by nmca »

Personal experience with this: I find a lot of AI articles to be low-quality, so I read the first paragraph, thought "whatever mate" and fast-scrolled looking for samples. Hit the bottom, read message, re-read critically.

So, I didn't think it was interesting, but it totally passed for blogspam for me!

Re: OpenAI's GPT-3 may be the biggest thing since Bitcoin

#400

Post by kristopolous »

visarga wrote: post #239
kristopolous wrote: post #115

It reads diffusive, unfocused and meandering, like a really bad writer who knows how to spell well and form proper sentences. All these AI generated systems have the same smell so far - from the ones that create art, music, literature - a convincing imitation of a lousy practitioner. This is fine for the background ... suspenseful music for a television show for instance or some descriptive balls-and-strikes style jo…

You're commenting on a rapidly evolving field as if it were static and we could imagine the results of the future models.

Everything I've seen so far points in one direction. I don't have a crystal ball, the future is unknowable, but I wouldn't be surprised if that pattern continues.

AI is kinda like a fake plant. They look pretty good, almost identical to the real thing, until you start feeling the leaves, looking for roots, you know, deeper inspection. Then you see it's all just plastic.

It's practical, useful, and makes life easier, great. It'll replace a lot of jobs, indeed. But in the same way you won't be able to harvest from a fake plant, there's no real insights in any of this generated material.

The fake plants of ai could unexpectedly start to grow fruit, sure. I wouldn't place any bets on it though.

Post reply on HN