blazespin

Well if you have a better business model, we're all ears.

Was OpenAI just supposed to live on the good graces of MSFT?

Do you know how much GPT4 costs? Plus training for GPT5? Plus supporting all those alignment efforts which just hinder the utility?

You don't call someone a liar because you have philosophical differences.

"Attention is all you need" was a silly name and most people who name their stuff are poking fun at it. Obviously, attention is not all you need. Just like lucene isn't all you need.

Yeah, they could have had the threat of eminent domain to force a deal, but they should have gone that route instead.

Saying something like "sell to us or we'll force you to via eminent domain", buying the land, and THEN complaining about the price tag is ... brutal.

Or some type of record where the farmers got into a smoky room and said 'let's all agree to jack up the price!'

That said, wow, this seems like a blatantly horrible hardball tactic or probably a result of sour grapes for someone who was a fool with their money.

I hope this gets dismissed with extreme prejudice and flannery has to pay legal fees.

bullshit is anything you don't need. you've been trained to think you need it in order to keep you and everyone else busy.

I've seen people who get promoted above their band, they are not happy campers.

I bet it really sucks, tbh. They did all this over promising and now the only way they can deliver is by grifting like this. That sounds really stressful to me.

It's a bit surprising how poorly DeepMind has lived up to their hype. But they're an OK lab, maybe a bit overly vain.

Yeah, this type of grifting is very proforma for Researchers. I used to stress about it, but realize it just sort of goes with the territory.

It's also worth noting that the paper is blindingly obvious and everyone started doing this a long time ago but didn't want to tip their cards.

And that's the real contribution here - Google is tipping their cards. We now have a rough baseline to compare our results against.

It doesn't presuppose anything other than that some people told him aliens.

I mean.. just accept it at face value. Stop reading into it.

It's so hilarious: "This can't be true, people would have exposed it, there's no way they could cover this up!"

...

"It reads just like so many claims like we have heard in the past"

I mean.... come on people, try harder.

One simple explanation is that Grusch's story is mostly true.

eg: There is a secret project, they have been doing reverse engineering, and they did keep it from congress for a very long time.

This is all the stuff that Grusch can probably prove himself without second hand reporting.

It's hilarious watching how HN upvotes insane theory instead of crazy theory. Wild times!

There is a very very simple explanation, which is that everything he says is true. There is a secret program, they did reverse engineer, they did withhold it from congress, and people did tell him that there is alien stuff.

Why make up silly make believe?

The guy isn't claiming 'oh i saw aliens'. He could simply be telling the absolute truth.

Why so many people told him there were aliens, I dunno, but who cares. Those people aren't the ones whistleblowing and so it doesn't really matter, does it?

I for one believe the guy, 100%

My only complaint is that perhaps he should probably be a bit more circumspect about what actually exists.

Good newsnation coverage as well with Mick West - https://www.youtube.com/watch?v=nDqfGkifPEA

Mick West suggests it's possible that people involved in recovering crashing satellites and reverse engineering may have let their imaginations run wild.

AARO (pentagon) is pushing back on this, but some folks suggest that AARO doesn't have clearance. eg: https://twitter.com/LtTimMcMillan/status/1664383272875376641

So basically, your rebuttal is that it's not this crazy thing, it's this other very crazy thing.

Ok...

Look, I'm as skeptical as anyone, but please watch your rebuttal.

When has there ever been a senior official with a significant reputation coming out like this?

That in itself is 'extraordinary' evidence.

Is this proof that life is out there? No, but if you're not updating your priors, you're not rational.

No, it's expert testimony. I can't believe I have to explain that on HN...

Well, did it remain secret? Seems like this has been openly recognized for quite awhile, but propaganda forces at play made it seems like you're a nutjob to believe it.

the issue with hacker news comments these days is people don't actually do any due diligence before posting. center for ai safety is 90% about present AI risks and this ai statement is just a one off thing.

To be fair, this paper has been made obsolete in its entirety with recent research. It's not really their fault, but folks need to start publishing faster as posters or something if they want to provide something relevant.

A better title, knowing what we now, might be "To outperform GPT4, do more than imitating"

What I find rather amusing is they spend the whole paper dismissing it as ineffective yet still feel the need to worry about the 'ethics' and 'legality'. They don't cite anything with regards to a discussion/evidence of either, of course, and looking at the authorship list I don't believe any of them are lawyers or ethics experts.

The breathtaking audacity of calling distilling GPT4 'stealing' when GPT4 trained on data it has no proprietary right to.

uhm, it sends your hostname, detailed OS information, and by default, your publicly routable ip address (on the packet). That's pretty malicious.

I don't think the title was deceptive.

Exactly, that's the trillion (quadrillion?) dollar question. Have we hit asymptotic reasoning capability or not?

There are interesting things to do with synthetic data, so stochastic parrot might not yet be hitting asymptote.

The dictionary definition of reasoning says nothing about how the thinking is done, only that it's sensible and logical, which is exactly what GPT4 is. Limited, yes, but it reasons.

The resolution is actually fairly simple. It's an incredibly brilliant stochastic parrot with some limited reasoning capabilities.

Some folks will try to say it cannot reason, but they are wrong, there is extensive proof of that.

The only question is how limited are its reasoning capabilities. After spending extensive time on openai/evals, having submitted 3 of my own, and doing a lot of tests, I would argue that an average person of average IQ could out think GPT4 - as long as the stochastic parrot aspect wasn't a factor.

The sequence of these two threads is just too perfect. Almost likely someone is trying to make a point.

And another - https://www.youtube.com/watch?v=MIfhunAwZew

These folks have been working on this for quite awhile.

Here's a similar project that was done awhile ago

https://github.com/qrdlgit/simbiotico

No, the opposite actually. GPT4 enabled this.