Agents attention cooled while economics rebounded

The Cognitive Revolution

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

that's sort of how you generalize from interpretability as like a factoring tool to interpretability as a tool for alignment.

we're going to have to continue to be very thoughtful in our, in like what guardrails we apply

The Cognitive Revolution

Pick Your Poison: Zvi Mowshowitz on the Unipolar/Multipolar AGI Dilemma, OpenFace & Pacing the ...

If you say that's aligned, then I don't care about the thing you're calling alignment.

That is aligned in the sense that you solved the alignment problem and got to do what you want to do, but also we're all super dead.

The Artificial Intelligence Show

#228: More Rogue AI Agents, AI Lab Staff Ask Washington to Pace Development, Continuing Battle Over Open Weights & OpenAI Previews Astra

We may be one turn away on frontier models models to where these researchers do no longer feel comfortable that they fully understand what the models are capable of and how to put guardrails in place to safely release them into the world.

The Cognitive Revolution

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

Obviously, one of the big kind of received wisdoms of AI product development over the last couple years has been try to build something that will really work for the— with the next generation of models, even if it doesn't quite work yet.

are we still in that regime? And if so, is there Are there things that you're like wanting Opus 5.1 to be able— I, I don't know if you're even able to use Fable, interestingly enough, but is there something where you're like, I want my Opus 5.1 wishlist, I, I really hope it cl...

AI For Humans

Seedance 2.5 & The New Age of AI Video (w/ Theoretically Media)

You do not look as good as that. Yeah. And then the other side of that, if there's any kind of like product branding in there, or if it's seeing it as, as a real person, or if any one of those supermodel dudes have any—

At the top line, I don't. Um, I'm going to use the best product, um, that provides the best results. Like, there's no, uh, uh, as we, as we, as we said in the '90s, um, uh, information knows no, knows no borders. So, you know, um, no, I'm, you know, I'm gonna jack into the old...

The AI Daily Brief

Google’s AI Leadership Shakeup: Disaster or Exactly What It Needs?

unbelievably powerful self-evident use case that for a ton of small business entrepreneurs who now make their living from Shopify shows just how valuable this technology can be

No Priors

Chasing Trillion-Dollar Companies, Founder Ambition, Token Budgets, and Regulatory Capture with Sarah & Elad

I think I'm looking at more, uh, both a little further out and then I'd say like, I don't know that there are that many markets that are gonna get to $100 billion of revenue in the next couple years that aren't Like inference, right? Tell me what else, what else you think in t...

Practical AI

Models, Harnesses, and Multi-Agent Systems

you can do a lot of what they do at much, much, much reduced costs and host it yourself and get the economics on your side

how do we get there at the most affordable cost for what it is that we need to accomplish? Which I think is a lot more sensible.

AI Inside

What Happens When AI Outsmarts Mathematicians

There's a price war going on here. Yeah, which is interesting. And, and, and in response, uh, OpenAI also lowered its— some of its prices. So I think we see a race toward the bottom here on pricing. I watched a video on TikTok today. I didn't put it in the rundown, um, from a ...

Last Week in AI

#253 - Opus 5, Gemini 3.6, Kimi K3, Hugging Face Hack

Let's now move on to Terminal Bench 2, Terminal Bench 3, and ultimately this. And there you go. So here we do know that this particular model, Opus 5, is outperforming all other models on a cost per task basis. And that's really where they're trying to differentiate it is cost...

The Cognitive Revolution

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

agents are also advancing research and interpretability in many ways

the sort of craft that goes into imbuing the right types of tastes and the right types of decision-making capabilities into any agent goes a super long way.

Eye on AI

Why People Are Paying 10x More for AI | Sid Sheth, d-Matrix

Do you see emerging companies, startups that are agent-first? The New York Times had a piece recently about a guy and his brother running a $1.6 billion sales company that's basically an agent-first company. Do you see that segment of the economy growing?

there's this iterative recursive loop, whereas AI gets smart, it learns to build tools that it can use in more efficient ways and more capable ways.

Practical AI

Models, Harnesses, and Multi-Agent Systems

That's where the autonomy of what agents are, the potential of what agents can do is really realized

The Cognitive Revolution

Thinking in Silico: Goodfire CTO Dan Balsam on Concept Manifolds & a $1000/Month ML Research Agent

I'm just always amazed when I think back to the kind of toy models of superposition is only like 3 years, right, ago now. And we have come so far.

where the geometry components come in is that the features aren't like, sort of like naive, maybe like SAE build take on things is like, you could imagine that the model is encoding like a bunch of totally orthogonal concepts to each other

Eye on AI

Why People Are Paying 10x More for AI | Sid Sheth, d-Matrix

The models have now understood that they have better capability. It's almost like a human being, right? Once you realize that you have acquired a new capability, you become more confident. And same thing with these models. They tend to have become more confident with disagreei...

No Priors

Chasing Trillion-Dollar Companies, Founder Ambition, Token Budgets, and Regulatory Capture with Sarah & Elad

I think the idea that the models can improve their own training if the leading scientists working on this believe it, and it's an extension of what we are already seeing in code and math, of course you should believe it, right?

The Cognitive Revolution

Pick Your Poison: Zvi Mowshowitz on the Unipolar/Multipolar AGI Dilemma, OpenFace & Pacing the ...

it obviously also reflects a major weakness in the models where they have this knowledge cutoff and they don't really know What's going on?

Eye on AI

Why People Are Paying 10x More for AI | Sid Sheth, d-Matrix

Wow. That's amazing. What about edge compute? It seems to me that this would apply the latency question. Are you moving at all into the edge where latency is so important?

We could. I mean, absolutely, we could. We are not currently. The platform, the way we have built it is a platform that is built with like a Lego block approach, right?

The AI Daily Brief

Why the Data Center Fight Has Little to Do With AI

there is so much opportunity to make this infrastructure work for everyone.

Gradient Dissent

40 Trillion Tokens a Day (Yes, More Than OpenAI) | Lin Qiao, CEO of Fireworks

KLD is a metrics of of the precision. So what it means is we reach bitwise equivalent. So the result in running from the training side, the result from inference side is bitwise equivalent. And why that matters is after you train, you need to deploy, right? So you don't want t...