Catloafdev

>the only reason ... is that current AI inference is unnecessarily RAM intensive

There's no indication that this will change, and every indication this will continue to grow. This is not some temporary thing. Demand is already exponentially higher than what is possible to produce, and there's no current reason to believe it has any ceiling.

I fear that, whenever consumer-facing supply finally starts to return, the opportunity cost will be priced in. Consumer RAM may end up being nearly as pricey as HBM for a while, and I don't see any reason why it would ever go back down on it's own, unless there are entirely new fab processes that don't meaningfully overlap HBM.

It's not strictly compute, because this has noticeably improved in open-weight models too, such as Gemma and Qwen. I suspect they noticed this issue and adjusted their training to be better about it over time.

No, it doesn't. Nobody would go do this just to report it to Sony for a few pennies. That is a non-existent motivation for people in the know, who understand the opportunity cost.

This seems extraordinarily sketchy.

I'm not saying "LLM's are only capable of AI slop."

We're specifically talking about the context of 'LLM-aided human output' not about general capability of these systems.

People can sniff out LLM generated stuff relatively quickly, the WR raises the floor.

The bar just took a very noticeable jump that people are still adjusting to.

Anything the WR can do on it's own is often seen as 'AI slop' - so now what matters is the QB. Because any random junk QB that doesn't meaningfully improve the WR's output is indistinguishable from AI slop. The distinguishing factor is solely the QB.

Sometimes I wonder what goes on in someone's brain to think these things.

Then I realize I'm probably better off not.

That's the verbiage racists use to try to make it sound like it's not racism.

Spoiler: it's racism.

No, it definitely was not as bad 10 years ago as it is today. Just objectively not true at all.

So if an OpenAI employee just went on his own and did this, rather than 'an AI', what legal repercussions would there be?

I wouldn't assume this administration gives a shit about inflation or rising interest rates or how the rest of the economy is doing. Do you have any reason to believe they do?

I don't think anyone really cares about 'the measured impact the data had on the exact result' - a question which is fundamentally difficult to answer accurately in the first place - but rather whether the data was used in training at all - which as Tristan described, was extensive, beyond simply a 'single chat.'

Can you explain the difficulty in engineering a search apparatus over a corpus of text data? Actually searching through it may not be easy, sure, but it's work that's doable, and creating an index is relatively trivial.

Frankly, I don't buy this difficulty argument.

They know which model was used to come up with that particular idea.

A text search over the corpus of user data used in the training set can only take so long.

It may not be easy, quick, or simple to figure that out - absolutely fair.

But it is knowable. Their entire business is built around training models - they have the ability to know exactly what was in any given training run.

I guess time will tell.

These responses seem to me to make it abundantly clear who's telling the truth here. I wonder who this fools.

It would be extraordinarily easy to simply say, this model was not trained on your work, if that were the case.

It's telling that they refuse to acknowledge the root issue here, and are attempting to shift the conversation elsewhere.

Ya, I'm not sure why anyone would choose to use this over any other option that doesn't sacrifice your privacy.

Google search has really gone to shit in this AI era.

The decline has been steep and noticeable. It really makes me wonder what happened behind the scenes over there.

Because it's the best option currently available.

It's really easy to shit on AI benchmarks, but that noise is useless unless you're offering a solution or a better benchmark.

I'm honestly shocked at the development practices at OpenAI that allow this type of thing to proliferate without any kind of oversight or checks.

I guess it's just "do whatever the hell you want" over there, huh?

Eh, I definitely can't agree with this one. Some of the most beloved games of all time had effectively no on-ramp of any kind - especially if you're willing to go back 30 years - and none of the fans of those games would say "they are not compelling experiences" because of a learning cliff.

You're totally allowed to have that opinion. Time is valuable. It does have a tutorial nowadays, but it's still not easy.

Not everyone _wants_ to learn Dwarf Fortress, and that's fine - it's not for everyone.

But I'm just providing context that, for those of us that have learned, it is a great game.

Yep, the game is difficult to get into, it's not for everyone.

The game is great, though. You are in the "I have no idea what's going on" phase.

I know HN sometimes has 'show' posts, but usually they aren't just ads, usually they have _something_ more than just a plain old "hey come spend money on my product" ad. Like a tech write-up, or something.

This is just an ad.

I appreciate the response, I should have finished reading through the whole thing first. My initial reaction assumed far less usage of AI to analyze the data.

Edit: I should have read through the whole thing first, ignore me

I don't get it, are you implying everyone who's not a fan of this change is a Google employee?

That's a higher percentage of the population than those who voted for the winning candidate in the last presidential election.

I wouldn't say that's comparable to ~15%.

Good point, I didn't see that. I guess I categorized them in the same bucket of 'runs on 128gb machines'

Guess Qwen 4 is the one to wait for.

Very curious to see how this compares to Deepseek v4 Flash. I have to assume they wouldn't be releasing this if it was worse.