This title shouldn't even be allowed for this research.

I see projects like this as both cool and terrifying for the economy

To me, all signs point to US over investment in hardware capacity to support the "frontier" when in reality the next major breakthroughs are likely radical optimizations that end in increased consumer training and inference efficiency.

I'm not sure how that, combined with open models inching up on frontier intelligence, is compatible with these massive AI IPO valuations?

It seems like closed model AI demand has a lot of contingencies that are ultimately fragile and possibly short sighted.

The article is about a guy leaving Microsoft and the disbanding of an internal team called "AI Frontiers".

Yeah, I'm speaking to "scaling vs research" and where investment is going. Which is relevant to the article. Accelerating scale requires demand, closed frontier demand is what I'm questioning.

And gp's comment is relevant to the particular problems the guy (John Langford) cites in his comment about leaving. IE, that research first winds being only about scaling and then winds-up ignored because "just scale" can be an easier answer.

If they make more efficient models won’t people just use the big data centres to make and run even bigger more powerful models than they can currently support? Or massive swarms of agents to look at all different avenues of a problem?

Do you think humanity is just going to run out of problems that need the smartest intelligence?

If so, then yeah, increasing efficiency 100x means we hit that point much faster.

If not, then it just means we make improvements 100x faster.

Everyone wants radically more efficient AI. That doesn’t mean it’s coming any time soon. And when models are twice as efficient, guess what happens next. We try making one twice as big for the same cost.

Anyone have a non paywall link?

Are these models harder to reason about given they produce tokens non-linearly? Especially since we rely on non-latent reasoning to know what the model is "thinking".

So like... do I get access?

This is mostly to be expected.

After Deno did layoffs we saw a consistent decline in announcements and innovation.

Sad day, but unsurprising.

It feels like the only thing left people are building are harnesses? Harnesses of harnesses?

Harnesses are only useful when frontier models are incapable of designing their own efficient interfaces with systems which they are improving at rapidly. This will be seen as a transitional artifact of a specific time in the development of general intellegent systems

I'm playing around with it for my MCP hiring protocol ojcp[1] and it seems to work very well for signing attestations at the header level.

[1] https://github.com/ojcp-org/ojcp

Thanks for sharing, it is interesting use case

It might soon be implemented into https://github.com/dakheera47/Job-Ops so yes, indirectly