BenzeneDream

When pigs fly, of course

You actually think recent chats were used in training of these models to THEN complete the proofs? This seems incredibly far-fetched and almost impossible based on how LLMs train. Occam's Razor says that simply the models are getting increasingly more capable in these domains.

This seems extremely unlikely to me, just more wishful thinking. To think there is going to be some other amazing technology beyond AI in a few years... hmm I sincerely doubt that.

"that is overly represented on internet forums like here. "

You don't think the recent successes of Anthropic and OpenAI point to this? There are plenty more people using AI agents than that are on HN, surely you must know that?

You've been hearing for years now that the capability in the last 2 months has skyrocketed? I find that hard to believe considering the start of this year I was still copying and pasting from ChatGPT...

Sonnet 4.5? Doesn't compare...

"With an AI bubble implosion now all but inevitable" 2023 is calling.

There is no bubble. There is no wall.

This just reads like cope, like you WANT them to be failing rather than the actual truth. Shutting down SORA proves nothing. How many products and services has Google shut down?

Video is costly and wasn't going to get them to RSI like math and coding will. While I enjoyed Sora and it showed just how far AI has come, their decision makes sense.

And to think just how amazing agents are at coding compared to even 6 months ago... I'm not complaining!

What are you talking about? You provided no evidence to your claim that you refute how good their 'inference' margins are.

It is WELL known that the margins on inference are extreme across the board, something like 95%.

For now. But the trend remains that bigger, smarter, general purpose AI beats smaller specific purpose models. This is why GPT 5 and 6 remain better at OCR than anything Mistral has put out.

You need to travel more if you think Uber and Airbnb are a "scam". To be abroad in certain countries and NOT use Uber or Airbnb is potentially putting your life at risk, especially getting into random taxis.

So... maybe there is an actual reason many of these companies had/have success that you may simply not be privy to?

What happens when you zoom in to a human-made painting and notice an error?

The video demo is epic!

Gimp is terrible though and they've had ages to change their UI/UX but they won't.

To pay for the training of the models which are getting bigger and more expensive. So intelligence is getting cheaper overall but the need for ever-increasing intelligence can't be sated.'

That's clearly a voiced bilabial fricative... so like a soft /b/ sound!

And humans don't iterate? Its a matter of time before there is nothing a human can do better than an AI in programming or math, let alone other fields.

You write with such confidence like this is an objective opinion, and yet one I completely disagree with. Confident in information which is false? What.. ChatGPT 3.5?

Of course, that will last for a time, until it won't.

I don't see a reason why expert humans will remain more expert than AIs.

"If you really cared about it, the code would not have stopped you doing it by hand." This is insane to me. So you have unlimited time?

Definitely not. Look at what Deepseek costs. AI companies spend a fraction on inference, it is not expensive for them to run these models like it is to train them. Coding by hand is not coming back.

The glasses are great! Considering they have a light that shows when you're recording and it is NOT easy to disable, I definitely don't think they are 'pervert' glasses or spy glasses.

There are certain recordings where glasses make the perfect form factor.

I think most people would not continue to code if their coding agents stopped working, just like if the internet went down.

I agree its not on the same level as the internet/electricity but its been quite a lot more impactful than you are giving it credit for, and that has only been increasing over time.

Open source models are just going to... disappear?

No. The vast majority of SWE are not going back to 'hand coding' so this will not happen.

Totally disagree. Newer models are doing much more advanced things in domains like math, programming, and cybersecurity.

I don't see why a company charging more for less is a surprise or would lead to thinking that AI models have reached their limits. Thats an incredible claim and it would need incredible evidence. Astra is much more capable than Luna. Luna being 'good enough' for most of my daily work just shows how strong Luna is, not that Astra isn't much more capable.

As models get more and more capable, they will only appear to be at the extremes. You'll need to see them solving more Millenium problems to notice a difference as the tide is continuously raised for all models as well.

They are already turning profits and inference has shown to be a cash cow. And they've already secured compute for the next several years.

Why would it freeze if they can create more training material. Its already been shown synthetic training data is just as good if not better.

I've had access to Gemini's video models for awhile now and have done nothing with them. So who cares?

Its not pretty bad for people.

So you are upset at the fact that Codex resets usage more often than any other provider?

Do you think all the comments in here are positive about the model? Because they aren't. In no way does it seem astroturfed. And yeah, pretty boring to read that kind of comment every time.

Anywhere in the world I vastly prefer an Uber over a getting a cab off the street. Safer and overall better experience.