So what’s the tldr? :)
Not public
Does some things better
Rhetoric of AGI being tossed around, redefining the term
GPT-6 Astra
811–820 of 1001 posts
Re: GPT-6 Astra
#812The ARC-AGI-3 scorecard is extremely misleading given that it clearly states itself that "with [the responses API] harness, we estimate Sol would score in the ballpark of ~30%." but it shows a score of 7.8% for GPT-5.6 Sol presumably since if they updated the percentage for GPT-5.6 Sol to the score it would receive with the responses API harness they used for GPT-6 Astra they'd have to do the same for the percentage…
I've personally been facing this lately 5.6 at max effort and fable have done tasks for me that I previously that would be a nearly 6mo project and it took me a week. It also did it better than I would have. The task was to build a high performance classification model. It not only helped make an entire data capture pipeline but also made the sythetic data basline needed. Then it proceeded to build and test 100 diffe…
> If it was entirely up to fable max or sol max the result would have been pretty bad.
How can you know it will have failed? I don't think it's that hard, if you clearly define the goal well, and have a bit more compute available, and do some intermediary bookkeeping.
Re: GPT-6 Astra
#813anyone read footnote 7 for a bench, where it says something about musical quartet 2023??
idk man until I test it, I won't get excited.
Re: GPT-6 Astra
#814...as always, in Europe cool stuff is available 10 years later.
I bet my ** GPT-6 will be one of those stuff.
Re: GPT-6 Astra
#815I have nothing to say about the actual model, but unrelated--why do so many of these demos include people buying things autonomously? Even if I did trust an AI to get everything right, it's not like the AI can read my mind. If I was ordering food normally and without AI, I would want more control over the process--looking over the options, prices, thinking about what I really want. People don't know what they really…
Despite access to """"""AGI""""""" all the marketing teams at these companies can only dream up 2 things, buying plane tickets and online shopping autonomously. Sometimes they're feeling extra spicy and throw in sorting emails or something along those lines. I suspect it's because it's tailored towards VCs and other similar rich ghouls as a replacement for their overworked and underpaid secretaries
And all of these things can easily go wrong with this halfbaked "AI" technology - getting thow wrong tickets you can't cancel, loosing all your money due to bot buying random titanium cubes & deleting your emails!
Re: GPT-6 Astra
#816I have nothing to say about the actual model, but unrelated--why do so many of these demos include people buying things autonomously? Even if I did trust an AI to get everything right, it's not like the AI can read my mind. If I was ordering food normally and without AI, I would want more control over the process--looking over the options, prices, thinking about what I really want. People don't know what they really…
The core or the problem is also what you are describing was bought up by Yuval Harari in his interview with the economist.
To paraphrase some of his lines: We make decisions using our emotions and our thoughts. What makes us different from the AI is that we can be afraid.
To portray a guy as ordering "beef bulgogi", in the same breath as "email this rocket design marketing", while it _might_ seem appealing and resolute, though oddly fast paced, seems pretty ignorant of the human _quality_ that make most practical decisions messy.
Re: GPT-6 Astra
#817It's fun, but every new model release makes me even less interested to create cool stuff. Like, what's the point, if the next AI can do it in 5 seconds?
This has not been my experience so far.
That's good to hear.
How long and much have you been using LLMs for?
I think it takes ~1 year of heavy usage for this feeling to set in, it's like the 5 stages of grief, there's different phases.
Now a big overwhelm come from the fact that the AI can work a lot faster than I can verify/check manually. Combined with the fact that not using the 100% usage of the subscription feels like wasting money/potential, leads to a feeling that I should always be doing more, where at the same time, it's getting hard to remember what am I even doing, when working on 3-4 projects at once.
And I do work in multiple projects at once, because that's how AI works best nowadays, if you have two threads running for the same project, it gets buggy or wasted tokens (like two people working in the same file at the same time, merge conflicts ,etc.)
So the solution is to have the AI work on many, independent tasks at once, but it gets overwhelming for a single person to keep track of all...
Re: GPT-6 Astra
#818It's fun, but every new model release makes me even less interested to create cool stuff. Like, what's the point, if the next AI can do it in 5 seconds?
You're not being ambitious enough! Spend your tokens now building the primitives and foundations of much larger, complex systems. No matter how much faster and more efficient models get, eliminating the gruntwork will always pay dividends.
That's what I am doing and usually telling others to: don't prompt the AI to do something, prompt the AI to create a system that does that something.
It's true, it's fun to have those systems, maybe I care more about them than the product, because the promise is "better, cheaper results in the future". But those systems are software too, and only tools, and will likely too be obsolete in 1 year from now.
I think it's a bit better when creating stuff now with a very short-term view: create something that is useful NOW, it doesn't matter it won't be useful tomorrow. Create it now and get it into the hands of people now.
Re: GPT-6 Astra
#819It's fun, but every new model release makes me even less interested to create cool stuff. Like, what's the point, if the next AI can do it in 5 seconds?
lol have you actually tried to build anything useful e2e? Leaving the AI to itself gives horrendous results.
Yes, many projects, used in production at various scales.
Some even for large clients, and they work well in production, clients are happy, yet I feel no connection to the work. If the client says something, I just copy-paste it to the AI and it fixes it/implements it.
There is too much code and complexity to check if the changes are ok, apart from seeing it if it seems visually ok and if the problem is fixed live. It almost always works, but it definitely doesn't have the same feeling as thinking of the problem, finding a solution, and implementing it, part which most developers enjoy in their work.
Re: GPT-6 Astra
#820[dead]