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

GPT-6 Astra

openai.com

Post reply on HN
Print view View original

911–920 of 1001 posts

Re: GPT-6 Astra

#911

Post by intenex »

Toutouxc wrote: post #721
intenex wrote: post #247

The 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 am reasonably confident that there's essentially nothing that I am better than Fable at despite generally being substantively above average on human benchmarks I genuinely don’t understand how an adult can say this with a straight face. I can take any single of my hobbies, start a mildly advanced conversation with Fable about the hobby and, within 5-10 turns, get it to contradict itself about something fundamenta…

Can you share some examples of this as shared conversations? I’d genuinely love to see this in practice.

Re: GPT-6 Astra

#912

Post by crossroadsguy »

HAL3000 wrote: post #136

Finally, OpenAI has a Fable/Mythos class model. 5.6 Sol felt like 5.5 on steroids, probably just a different checkpoint with a lot more RL post training. I wouldn't be surprised if there are some conceptual similarities to the kind of latent reasoning Anthropic sees in claude's J-space, although those aren't the same thing. Recurrent/looped transformers themselves aren't a new concept, but it's interesting to finally…

> Canceling my Anthropic Max sub when this ships.

At this point, it reads like people are cancelling old ones and getting new subscriptions every two to three days, whenever a new ,model drops, and quite possibly by the end of the week they are back to the old provider while still having active subscriptions with at least two to three others. Interesting times.

Re: GPT-6 Astra

#913

Post by intenex »

123894893 wrote: post #789
intenex wrote: post #247

The 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'd be curious to hear takes on what would make you think Astra is yet to be AGI Give someone 10 remote employees for a few months, 5 of them human, 5 of them AI. After a few months, check to see if the humans (manager, other coworkers) can figure out who is AI and who isn't. Would that be sufficient? I'd have to think about it. But AGI is supposed have human level capabilities, so this would be a necessary prerequ…

Being as good as a human at everything isn’t the same as being able to masquerade convincingly as a human at everything.

A better benchmark would be seeing which cohort of employees the manager prefers employing after a few months.

Re: GPT-6 Astra

#914

Post by crossroadsguy »

billypilgrim wrote: post #401
XCSme wrote: post #332

It'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?

„The depressing thing about tennis is that no matter how good I get, I'll never be as good as a wall.“ -Mitch Hedberg

Well people will still pay to watch a human player play tennis even if a robot could play infinitely better than that human. Try something like that with an employer and software engineer combo (and no, you don't even have to imagine that).

Re: GPT-6 Astra

#915

Post by user43928 »

qsera wrote: post #740
intenex wrote: post #247

The 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…

>what would make you think Astra is yet to be AGI... Can it detect if I feed bullshit (by bullshit I mean stuff that contradicts with its own existing "knowledge") in its training data? If not, then I think it is a good indicator that it is not intelligent at all, let alone AGI... And I think discussions on whether these models are AGI or not are AI marketing triggered. And that is exactly what these statements are t…

Anthropic's J-Lens research indicates so.

Like when web results are fake it would show things like 'FAKE PROMPT INJECTION'.

Or in a safety evaluation with a contrived scenario it was 'FAKE FICTIONAL'.

Re: GPT-6 Astra

#916

Post by Dlemlo »

dakolli wrote: post #526

Why is everyone so excited to be replaced and become reliant on some billionaire's thinking machine? These are just going to be used to turn you into a rather dumb reliant paypig.

Its still human made frontier progress.

And for sure it has a tremendes amount of implications, but its not the fault of the technology (we found, not invented).

And i'm only living once, my main motivation is not to just live day in day out the same stuff, i'm quite happy to see progress.

Am i worried about the future of our planet? For sure.

Re: GPT-6 Astra

#917

Post by Dlemlo »

efavdb wrote: post #560

Seems like only yesterday that gpt 5 was supposed to mark our downfall

Its hard to understand how exactly these AI people mean it.

When i say "AI is changning the world" its more like "I can already see how this technology will continue to become better and better and has more impact every single day. It already affects people and it will have fundamentally changed A LOT in 3-15 years"

But lets be very realistic and clear: My computer systems at home were exploitable a lot more often this year than any year before JUST because of GPT or Claude.

Re: GPT-6 Astra

#919

Post by falcor84 »

ex-aws-dude wrote: post #536
astrobiased wrote: post #365

I can’t help but notice how much this echoes Francois Chollet’s On the Measure of Intelligence: https://arxiv.org/abs/1911.01547 Most of frontier-model progress still looks like skill acquisition optimization: broader benchmark coverage and performance, more domains absorbed into the training distribution, and increasingly strong performance within that surface area. It seems more about coverage-driven competence. So…

They can do new tasks with in-context learning but its obviously limited by context window

But what does it mean in practice? Obviously we humans also have a limited cognitive capacity.

Let me offer a thought experiment: Let's say that tomorrow we discover Atlantis, with a treasure trove of books about their culture and science, written in a dialect of ancient Greek that we know how to start to analyze, but no one can read fluently. And let's say that you are a billionaire really curious about their culture and want to converse with an "Atlantean expert" as soon as possible. Would you invest your money in a "we-hate-ai-slop(tm)" group of researchers who would abhor AI and instead delegate the books to a massive number of human grad students? Or in a small group of researchers who are willing to use AI agents to go over these? Or maybe just open a chat session with GPT-6 yourself immediately? What would most effectively assuage your curiosity?

Re: GPT-6 Astra

#920

Post by aizk »

atonse wrote: post #162
HAL3000 wrote: post #136

Finally, OpenAI has a Fable/Mythos class model. 5.6 Sol felt like 5.5 on steroids, probably just a different checkpoint with a lot more RL post training. I wouldn't be surprised if there are some conceptual similarities to the kind of latent reasoning Anthropic sees in claude's J-space, although those aren't the same thing. Recurrent/looped transformers themselves aren't a new concept, but it's interesting to finally…

yeah i'm wondering the same way... especially in light of the 20x debacle (where we found that 20x of Max vs 5x only applies to the 5hr limit, not the weekly limit, whereas OpenAI's 20x actually is 20x overall). Also Opus 5 has been really tough to work with. I can't understand half of what it says, it's just so damn obscure.

I still use Opus 4.8 for a lot of tasks because I can't stand the way it talks.
Post reply on HN