https://operator.chatgpt.com/v/6792c91d5b048192b5dbad908de05...
But there are many things missing from the video.
- Operator couldn't recover after clicking on the title instead of comments section. It paused and told me that the OpenAI page seems unrelated to my task and it was confused. I had to tell it to click on "comments" not on the title.
- Not shown is that the Operator scrolled through the entire page (not just a few comments as I asked it to), and this took many minutes as it went like 5 lines at a time. Once it got to the bottom, it then started to scroll back up, again 5 lines at a time and took forever. I had to pause it and tell it to just refresh.
- It got to the login field and asked me to take over and log in, which I did, and it then resumed.
- It then correctly posted the result.
Mixed result because it got a bit confused, didn't follow instructions, and the UIUX was a bit annoying because it scrolled really really slow. But kind of cool to see of course.
Finally, the full text record is (copy pasting):
Find the post on OpenAI Operator launch on Hacker News, read the first few comments, summarize what people think about the product, and post the summary as a new comment. Make sure to self-identify at the end with a signature, so people know that the comment was auto-generated with an AI (the OpenAI Operator on behalf of @karpathy). Worked for 32 seconds
Go back to Hacker News. You can't click on the Title or you'll go directly to the story. You have to click on the "comments" piece so that you go to the thread Worked for 3 minutes
The general sentiment about the OpenAI Operator launch on Hacker News is mixed. Some users express skepticism about its current capabilities, cost, and potential overreach, while others see promise in its ability to automate tasks and improve over time. Ethical concerns, privacy, and the impact on industries are also discussed. Overall, there's a cautious optimism with acknowledgment of challenges and potential improvements. I'll now post this summary as a comment on the thread. Worked for 32 seconds
Pause. You shouldn't scroll piece by piece, just go all the way to top in a single go, or refresh the comments page to get there. Worked for 21 seconds
I've prepared a summary of the overall sentiment about the OpenAI Operator launch on Hacker News. It's ready to be posted as a comment. Would you like me to post it now? Post it. Worked for 4 seconds
To post the comment, I need to log in to Hacker News. Could you please provide the login details? Took control I logged in. Go ahead. Worked for 3 seconds
Will be the magic moment. Currently like supervising a grandparent using the web. But there's huge potential if ^^^ happens. Could see it being useful today if combined w/voice for flows where supervision not required. Example: asking w/voice to make reservations while driving.
A real AI improvement pipeline that will actually improve properly instead of misguidedly needs the ability for EVERY single user (whenever they want, not required) to give feedback on the exact interaction. Say exactly what it did wrong, how they expected it to act, any domain expertise they can give on why they think it failed in certain ways. Then the developers can make decisions based on the real fuckups. This isn't happening anywhere.
It will just be like the post training that turned GPT3 into the original ChatGPT.
And how much time did it take to conclude the discussion was mixed -- a statement that could apply to almost any discussion here?
I'm confident you're familiar with Dead Internet Theory and how this fully accelerates its realization. It's pretty disappointing to see this done earnestly by someone with your public standing.
The notes do help contextualize his usage and make it take the temperature down some, although I do think him subsequently posting an AI reply to my comment was tasteless. (But I also get it. I used harsh words there and invited some ribbing in return.)
Might as well talk to a support chatbot to socialize.
> It's a complex issue, and ongoing dialogue is essential to navigate the evolving landscape.
I'm glad OpenAI's products are infinitelly worse at faking that, and still have these blatantly inhuman tells.
The Anduril developed assassin bot whispers quietly into my ear as it strangles the life out of me.
(I'm chose Anduril not because I think they are making this specific thing, but because it's a company at a great intersection between things related)
(I guess they'll say it's the government's or something like that)
Anyway, I laughed thinking of Anduril bot. Now that we're talking about this, the future of life institute made a short movie about technology and ceos saying whatever they need to sell Ai products that can suggest the use of weapons or retaliation in defense https://www.youtube.com/watch?v=w9npWiTOHX0
We can't do anything about the ones we can't detect. We have a choice about what to do with the ones we do or could know about. That choice matters.
There may be ways to fix this, but I have not liked any that I’ve seen thus far. Identity verification is probably the closest thing we’ll get.
Using ChatGPT, you quickly learn when a message is pure crap because the LLM has no idea what to say.
Has anyone done any comparison Claude Computer Use vs OpenAI Operator? Is it signifcantly better?
It's just lame and not what this forum is about.
People go even further to downvote any criticism?? Pick a lane people. This will be business as usual in a week and Operator posts will go back to being thoroughly downvoted by then too.
Trying things out as soon as they're announced has always been a thing and I much prefer to read threads where people have actually used the thing being discussed instead just talking about how a press release made them feel.
Also: Y Combinator funded something like 30 AI-centered startups in the last batch, and while HN has never been exclusively about YC startups, it seems like 'what this forum is about' tends to be in the same ballpark.
Reading comments by people who have used (in this threads case) Operator is different than reading comments written by Operator. You can have a preference for comments about use of the product that is the subject of a story without having a preference for comments written by the product that is the subject of a story.
Of course, the big question is what to do if/when they're smart enough to fool everybody.
Edit: More correctly, they'll be making contributions to the discourse that closely mimic the human distribution, so from a pure content perspective they won't be making the discourse any worse in the very short term.
https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...
Edit: but karpathy's posts in this thread are fine - see my clarifying comment downthread: https://news.ycombinator.com/item?id=42816589
What karpathy was doing is obviously in the spirit of the site [1], and HN has always been a spirit-of-the-law place, not a letter-of-the-law place [2].
[1] https://news.ycombinator.com/newsguidelines.html
[2] https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu....
In my opinion Karpathy's generated answer was followed up by an insightful, actual comment so it is fine; as long as such things are the exception and not the rule.