# AI Infrastructure Wars and Model Efficiency Shifts

**Podcast:** Last Week in AI
**Published:** 2026-03-03

## Transcript

We once again want to thank Box for sponsoring Last Week in AI.
If you're trying to transform your organization with AI, you're likely facing a common challenge.
Most AI tools are great at public knowledge, but they don't actually know your business, your product roadmaps, your sales materials, your HR policies, the content that actually makes your company run.
And that's where Box comes in.
Box is building the intelligent content management platform for the AI era, serving as a secure, essential context layer for Box AI agents to access the unique institutional knowledge that makes a company run.
And that's a key idea.
The power of AI doesn't come from a model alone.
It comes from giving AI access to the right enterprise content.
And that's what Box does.
It goes beyond file storage by connecting content to people, apps, and AI agents so teams can turn information into action.
With tools like Box Agent, Box Extract, Box Hubs, and more, organizations can accelerate knowledge work, pull intelligence from unstructured content, and automate workflows.
So if you're thinking seriously about your company's AI transformation, think beyond the model.
Your business lives in your content, and Box helps you bring that content securely into the AI era.
Learn more at box.com slash AI.
Hello and welcome to the Last Week in AI podcast where you can hear us chat about what's going on with AI.
As usual in this episode, we will summarize and discuss some of last week's most interesting AI news.
You can also check out our Last Week in AI newsletter at lastweekin.ai for articles we will not be covering in this episode.
I am one of your regular hosts, Andrei Kerenkov.
My background is of having studied AI.
and a PhD and now working at the startup Astrocade.
And I'm your other co-host, Jeremy Harris from Gladstone AI.
I'm going to do AI national security things.
So we missed a week and then we got hit with a week with tons of papers, tons of announcements.
This is just going to be a giant episode, I guess.
That's just where this is headed.
Yes, and we're going to...
Try to be efficient and not make this two hours, but we'll see how it goes.
Usually when we get into a paper section is where things start being a bit longer.
We'll see.
And you mentioned Glassstone.
I do, as always, want to apologize for missing a week in this last week in AI podcast, but it's all about day jobs.
I also couldn't kind of make it up.
Fun fact, I guess I don't think I've shared it.
Our startup just raised our Series B recently.
Oh.
So it's a busy time.
There's a lot to do.
Is there an option?
No, we aren't publicizing it or anything, but it's not secret.
So I'm just going to save that much.
Very cool.
Congratulations.
That's great.
Yes.
So it's fun.
And if you happen to be in the Bay Area and looking for a job, we are hiring engineers, back-end, front-end, growth marketing.
So you can feel free to email me if you want.
Amazing.
And now before we get to the news real quick, we do have a lot of listener comments that I want to acknowledge.
A couple new reviews on Apple Podcasts.
I remember we called out a review that said we are mostly back in 2026.
When you're back now, there was an update that says totally back at 100%, which is good to hear.
We're still missing weeks, so I don't know if it's 100%, but it's more, it's the same as in 2025.
And a couple more that I appreciate technically and philosophically excellent.
I don't know if anyone has said ever we are.
philosophically excellent but I appreciate that and along that note I guess in response to us calling it out briefly a lot of people did comment on the handling of politics and the state of politics in the US in particular and appreciative of being direct about it which is what I tend to do Jeremy given your job national security maybe cannot be as opinionated or direct but We will, as always, try to keep our politics related to AI, which this week, as you may know, we'll get into that a little bit.
Yeah, if anybody wants to see me squirm, this is going to be the week.
Yeah, it's not like the military or anything is related to what you work on.
I wouldn't imagine it.
So with that being said, let's get to the news.
Starting with tools and apps, and we're going to catch up on some stories from last week we didn't get to cover.
First up, we've got Sonnet 4.6.
So Anthropic released Opus 4.6 pretty recently, maybe even just a month ago.
Now we've got Sonnet 4.6.
Very similar announcement in that it's just a 0.1 bump from Sonnet 4.5, the previous model, but...
It's an impressive improvement kind of all around.
Another increase in the contact size up to 1 million, which is a very big deal if you're using these in production.
And similar to the Opus update, it comes really just a couple months.
I think the last one was November, if I remember correctly.
So it's a big improvement in a relatively short time span.
In 2025, the time between these models, to me, felt longer.
And the jumps didn't feel as steep.
So I think we're seeing a lot of indications that post-training, reinforcement learning, kind of scaled up reinforcement learning where you don't need new data, maybe kind of allowing continuous training and continuous improvement, at least right now, is my feeling.
Absolutely.
And also distillation, right?
So as we're talking about the performance of Opus, the better anthropic gets, the better labs get at distillation, the more that translates into models like Sonnet getting better faster, right?
So the difference, the distance between Opus and Sonnet might close as you get better at distillation.
It's not like they're necessarily using Opus, let's say only or...
to the exclusion of other models, but because there may be even bigger models in-house, right?
Opus itself will be a distillate of some even bigger model because typically you don't actually serve the huge mega model that you train.
But anyway, yeah, so distillation is important here.
We'll be talking about distillation in other contexts too, actually later in this episode.
RKGI 2, 60.4% on that benchmark.
Wow, that happened fast.
This is not...
technically SOTA overall, but certainly in its weight class, so to speak, if we can speak of weight class because we don't know the parameter counts, this does seem to be pretty close to field leading, if not just field leading.
It's still behind, obviously, Opus 4.6, but also Gemini 3D Think and one kind of refined version of GPT 5.2.
So this is an impressive model.
I would expect this is going to be the model that a lot of people use day-to-day encoding.
It's really impressive.
I mean, having played with it over the last couple of days, This is a genuinely impressive model.
And Anthropic is on fire.
Anthropic is on fire.
And speaking of Arc AGI 2, we've been mentioning more and more, and we will mention it with the next story.
So perhaps worth giving a quick primer on the benchmark itself.
The reason it's called Arc AGI, well, I don't know the Arc part, but the AGI part is meant to be evaluating whether AI is human level in this case.
Human level meaning it's basically an acute test given sort of a set of patterns to complete or some sort of like problem that requires you to generalize on the spot given just a few data points.
So you can't like call up a fact, for instance, or find a solution from the web, something like that.
The models are meant to match humans where humans kind of intuitively are quite good at a lot of these problems.
So with RKGI, you're kind of the threshold of a bar.
There's a couple different ways to kind of win.
One of them is to give in limited compute and limited data to then get a score that is very strong.
And this was quite challenging for LLMs for a long time, I think partially because a lot of the problems are quite visual.
So if you try to explain them in text, they are very high dimensional problems.
There's like grids of pixels, right?
And that just doesn't lend itself well to LLMs.
So LLMs are getting better at multimodal reasoning and performance, which we don't call out as much, but is very significant, actually, in multimodal, given that now we want more and more work in our computers, being agentic, et cetera, et cetera.
So as a benchmark, it's quite interesting in that it's not kind of task-oriented on programming or QA or search or anything like that.
It's kind of raw IQ.
Yeah, it's also, well, I guess one way to think of it, an example, and this is a toy example to give you a flavor of what these evals look like.
It's almost like you have a model that might have been exposed to connect four in the prompt.
And then you're like, okay, now do connect five, right?
Play a game of connect five or you do regular tic-tac-toe.
Okay, now we're going to do tic-tac-toe on an infinitely large grid or something like this, where it's, you're trying to test explicitly the out of distribution generalization capabilities model that go beyond whatever it's been exposed to in the prompt.
All kinds of issues here around task leakage, as you can imagine, right?
And with IQ tests, if you saw the IQ test before, like last year's version or whatever, so you do get benchmark saturation from that.
But famously, RKGI 1 is now kind of the solved problem.
And so RKGI 2, I think François Chaudet, who is the researcher who sort of invented the RKGI benchmarks and who's done a whole bunch of things around deep learning frameworks in the past and blah, blah, blah.
I think he said something like he expects to do...
up to Arc AGI 4 maybe before we get to ASI.
I think that was it.
So he's not thinking of Arc AGI 2 as like the thing that once you get it, you get ASI.
But he sees it as a big waypoint, which is significant because he's a historically an ASI, long timelines guy.
That's right.
And Arc AGI 3 is coming.
Actually, it's announced for March.
Just looked it up.
Arc, abstraction and reasoning corpus.
And if you go, there's actually a paper on the measure of intelligence.
that he wrote and it explains it pretty well.
The focus here is on skill acquisition, efficiency, scope, generalization, priors, et cetera.
So it's one of the more useful benchmarks, partially because of its design and partially because from the very beginning, or at least very early on, if I remember, there was a held out test set that was private.
And to get a score, you need to submit your model.
And it's a bit trickier basically to see.
answers.
You're given some data as an example, but the whole point is you generalize from not much data.
And just one last piece of speculation I'll throw out there for Summit 4.6 and Proopic in general.
I have to wonder how much that cloud code data is helping them out because they're getting a lot of data to train on and it's hard.
It's easy to forget, but that's new data that no one else has.
Probably unimagined already part of a training here.
I'm trying to remember, I think Mario was on Dorkesh's podcast at one point fairly recently talking about why we haven't seen a kind of liftoff where whoever's number one in LLMs just like keeps compounding and running away with it.
Because you would think that that would happen, right?
These companies are dogfooding their own LLMs.
So theoretically, right, that should happen.
And one of the reasons that he said that's not happening yet is the uplift you get from these models.
Like a year ago, he said it was maybe like 5% uplift.
Now it's maybe more like 20%, which seems to imply that then we may be in for a phase transition fairly soon because 20% is quite significant.
So maybe his argument would be actually that what you just said is right on the nose, which would be quite interesting.
And the whole trick now, I think part of the reason we don't feel like we're jumping as much is that now more and more of the actual challenge is long-term.
task you know agentic task solving we're not getting to that super intelligence and i am always skeptical of the notion of super intelligence where you like immediately are some high brainiac that knows the answer the challenge and i think what these models will keep getting better at is actually working for four hours eight hours you know whatever without losing their mind yeah yeah And moving on to the next impressive model release from last week.
This has been a crazy few weeks for model updates.
Google has rolled out a Gemini 3.1 Pro.
I think they have...
Previously kind of previewed it, but now we actually have the rollout.
They announced the performance and this one gets at 77.1 on Arc AGI 2 compared to Gemini 3 Pro's 31.1%.
That's the big headline here is on that Arc AGI benchmark.
It is surprising to a lot of people that we are getting to that level of performance.
And that was also a bit...
taking aback by it.
But yeah, all accounts similar to Sonic 4.6 and Opus 4.6, a pretty impressive leap for just a 0.1 bump in the version.
Yeah.
Also, all kinds of interactive capabilities.
It can generate, in one case, it shows a 3D interactive starling murmuration, which if you didn't know the collective noun for starlings, it is a murmuration.
And with a dynamic soundscape too, so a match to audio and all that.
So, you know, a whole bunch of interesting capabilities.
I think pretty consistently we've seen Google go kind of multimodal as their differentiator.
You know, Claude obviously better on the coding side and then OpenAI going kind of more full consumer.
But yeah, it's kind of interesting to see them keep leaning into that, which is consistent too with, I think, internally, you know, when you think about Google, when you think about Google DeepMind, the idea of...
world model simulations as ways to generate training data for agents has just been a bigger thing for them typically.
And so I think you're kind of seeing that reflected in that orientation as well.
But yeah, big, big deal.
And that multimodality, by the way, is helpful explicitly on the ArcGi2 benchmark.
A lot of those benchmarks involve visual problem solving, looking at puzzles, looking at things.
So maybe that multimodality is an asset in that context specifically.
In fact, I would expect it to be.
So you've got both the reasoning and the sort of multimodality allows you to interpret and understand what you're reasoning about.
And one thing we haven't covered as much, but as a bit of a reminder, one of the strengths that Google has is on the pricing side.
Relative to other models, especially Claude, it is quite affordable.
You pay $2 for a million input tokens and $12 for a million output tokens, at least on shorter prompts that are less than 200,000 tokens.
Claude Opus 4.6 is $5 for that input and $25 for that output.
So twice as expensive, basically.
And Anthropic has sort of gotten away with pricing at a premium for a while.
They have seen that enterprise is willing to pay for the best.
They're pretty price insensitive.
But as kind of the models start being more and more similar, I don't know if cloud might be in a bit of trouble if they continue to be priced higher to this extent from the other models.
Yeah, I mean, for a lot of workflows, especially when you're looking at coding, I think people are willing to pay top dollar for just like whatever the best model is.
But yeah, you're right.
I mean, presumably we'll hit a point where that tired comment that I keep making about image generation models starts to apply more and more to code where it's like, For 80% of use cases, it's not a big deal which model you choose.
So we may be headed for commoditization along some axes.
And you got one more model release, not quite as big a deal, but I think also worth mentioning.
ROK 4.20 is in public beta.
This one, not quite as big.
In fact, I don't know that they release benchmark numbers.
What people on Twitter seem to indicate is it's not even clear if this is a new model or if they just tweaked the inference to make a model have multiple personas talking to each other and then synthesizing a single output, similar to Grok Heavy, where they ran multiple things in parallel and then synthesized a better overall result.
According to Elon Musk, ROC 4.2 will be about an order of magnitude smarter and faster than ROC 4.
I don't think that's true.
I don't think we're going to see a 10x improvement from 4 to 4.2.
ROC 4 already quite capable of not leading the pack anymore.
So yeah, still getting releases from XAI.
I'd be curious to see how they keep up given now they're being folded into SpaceX and a bunch of people.
I don't know if you covered that on here, but some of the co-founders and apparently a lot of the technical staff were transitioned out as XAI was folded into SpaceX.
Yeah.
So one of the interesting things about this release, I don't know if we're supposed to call it 420 or 4.20 or what the thing is here.
Let's just go 4.2.
We get the 420 joke.
We get it.
It's very funny.
I'm looking forward to Grok 6.9.
Apparently, the idea here is that 4.20, 4.20, is being framed more as this high-level expert tool, right?
So instead of just focusing on, oh, this is an edgy model that'll tell you things uncensored, which has been typically the personality of Grok or the kind of main focus, here it's more on these real-world concrete capabilities in medicine and engineering.
At least that's a lot of what this announcement is kind of highlighting.
It's interesting.
So the pitch here that Elon has is that you can just take a picture of your medical data or upload the file, get a second opinion from Grok, which liability, liability, liability, and all that I'm sure is being covered.
But this is basically the pitch.
So quite interesting.
It's a different twist.
You know, if this becomes a persistent thing, then this is kind of like Grok trying to carve out a maybe more monetizable corner of the LLM landscape and market.
And to be clear, this was just released in public beta.
There was no kind of big model rollout yet.
This is being tested out by some people on X.
But people on X have kind of been showing, and I don't know if this is easy to see or if you just see this in the thinking traces, but there are these four agents, Grock, Harper, Benjamin, and Lucas, who debate internally, fact check each other.
and help each other kind of get things right.
And that's why I say it's unclear if this is a new model or if it's a new inference paradigm or if it's both.
We don't know.
Next, on to some more kind of application side of things, Anthopic released a mobile version of Cloud Code called Remote Control.
So if you have a cloud mobile app, you can have basically an online session of Cloud Code that you can remotely hit up and people have compared this to OpenClaw where it's like a thing, an agent that lives out there.
It just hangs out and waits for you to reach out and ask it to do something and then it goes off and does it for you.
You've seen this before with Codex from OpenAI.
They actually very early on decided to do the online agent strategy where you reach out and kind of ping it and the agent goes off and does stuff for you.
And hopefully you get it right.
You come back later.
So here it's just easier for Claude to be used that way.
I'm kind of still a skeptic of that being useful right now, at least for more complex software engineering.
But it's the way of a feature, seemingly.
Yeah.
And there's, you know, a lot has been said about the security layer here and the security implications.
Obviously, you know, OpenClaw, your computer is just like its giant playpen, right?
It'll go bananas.
It'll delete files.
It'll, you know, it'll send wire transfers, whatever you want.
Whereas here, so when you actually run a command using Cloud Remote Control, it's set up so that you're basically your machine sets up an outbound connection to Anthropics API.
you're not actually like opening any inbound ports.
So your computer isn't exposed to the open web.
You're just basically pulling the API for instructions and maintaining that separation.
You can kind of have a remote window if you want to look at the process that's still running, but it creates that separation very intentionally.
This does reflect Anthropik's philosophy on all this stuff, which has consistently been, hey, you know, maybe we shouldn't fully trust these increasingly intelligent conniving agents to just like...
run roughshod on our computers.
So anyway, there you go.
Yeah.
And one thing to touch on here, just to be clear, it's also a bit different from Codex and previous, a bit different from OpaClaw as well, in that one way to use it is you just start a Cloud Code session on your local development environment, just as you do normally.
And then it just lets you talk to it via your phone.
So one of the very annoying things with these cloud agents is you have to then transfer all your files to the internet, to a repository, and they have to work in this little isolated thing where it's not your computer.
One of the ways this could be more useful is you just work in the same environment where you interactively work with cloud and you can pick up or check for updates when you go out for a walk, which I actually could see being useful.
And on to one last product update.
Perplexity has announced Computer, an AI agent that assigns work to other agents.
So the key here is that it coordinates multiple agents to execute user-assigned tasks.
The claim is, or the intent here is that it can run for extended hours, from hours apparently to...
months depending on what you want to do.
So it can do things like creating a plan for a marketing campaign or to build an app.
This model will then break it down into subtasks and assign it to specialized AI agents.
Interesting to see perplexity doing this.
It seems like we might be trying to find new avenues to be able to monetize given they've been doing search.
since the beginning and they have done deep research.
And now this is outside of that.
This is getting more into vibe coding and agentic vibe coding, which is not so much their real house.
Yeah, it's also exactly what you have to do if you aren't in the business or at the scale where you can train your own models, which perplexity is not, right?
They're sort of in this gray area where they have to aggregate, find ways to add value using other models.
This is that kind of play, right?
You have an agent that assigns work to other AI agents.
Those agents could be Gemini powered or Cloud powered or whatever.
So it kind of is this integration play, this platform play that perplexity is strategically in a natural position for.
Some might say it's their only option.
Like they have to find a way to position themselves as a way for you to easily kind of flip back and forth, kind of open router style almost.
I think this is a natural move, though, as you say, it is well outside the remit of the kind of deep research and search function.
It's a different use case, a different, I guess, habit stack that they're targeting, which will be interesting to see if they can break into it.
Computer, by the way, makes me think as a Star Trek fan that that's what they're going after, right?
Computer, do this thing, right?
Anyway, that just kind of seems like part of the goal.
So it'll be interesting.
We'll see how the product grows if it does.
Speaking of that, another fun fact, computer originally referring to people, right?
Computing up until probably the 50s, we had computers employed for like NASA or whatever.
Then that changed, obviously.
But now we're going back.
Now you have AI being computers, which is kind of funny.
That's right.
On to applications and business.
First, we've got Meta has talked to AMD.
They have a deal where they're going to spend up to $100 billion on chips over multiple years using AMD's MI540 GPUs and CPUs and perhaps other chips over time.
Interesting or...
Kind of notable.
AMD, of course, is competing with NVIDIA on this front.
They are not as present in the ecosystem for things like AI training, AI inference, but they do seem to provide a pretty decent offering.
We've seen them partnering with other companies.
If I remember correctly, OpenAI probably made a deal.
They made a deal with everyone.
So as everyone competes for compute and Meta is...
deciding to spend a jillion dollars on data centers, I suppose not too surprising to see us.
Yeah, absolutely.
So there's a lot that's interesting about this deal.
One of which, one thing is the sort of equity and warrant structure here that is actually in some sense the kind of main story, right?
So it's not that NVIDIA is just going to sell chips to Meta.
It's something like it's giving Meta like a 10% stake in AMD, you know, warrants for 160 million shares at one penny a piece.
but it's contingent on performance, right?
So there's the final tranche that comes in that requires AMD stock to hit $600 a share, which is more than triple its current price.
So that's a really big expectation.
So there's a lot of incentive alignment happening there between Meta and AMD.
This really is about essentially Meta trying to help AMD be a successful competitor to NVIDIA.
And obviously a ton of interest that Meta has in that outcome, you know, diversification from a geopolitical standpoint.
but also Meta itself has its own in-house chip effort.
So the multiple supplier approach, you're seeing that increasingly getting used.
You know, this is just a huge deal.
That's the other thing to flag about this.
$100 billion in chips, $600 billion committed to data centers just over the next couple of years, $135 billion in CapEx this year alone.
This is a really, really big, big build-out, and they're basically betting that the cost of not having Frontier AI infrastructure is...
just so big that they have no choice, right?
So six gigawatts of power is huge, right?
That's six nuclear reactors, right?
That's what we're talking about here.
So really, really big story on so many different levels.
I think this is Meta's big play, at least.
Let's say, at least for the week, they'll probably calm down.
We'll see what happens next week.
Right.
And as before, the justification is that Meta is working on their personal super intelligence.
So slightly different.
In framing from before, they are investing because they essentially want to take the lead or, I guess, be in the race for superintelligence.
Questionable whether Meta should be competing on that front.
We haven't seen anything come out from them recently on the model front ever since Llama 4, which was, what, almost a year ago now.
Ever since they didn't acquire a scale AI.
Ever since, yeah.
So you've got to wonder.
if they're gonna actually put all this compute to good use yeah we'll be looking out and seeing i want some gossip you know and speaking of nvidia next story nvidia challenger met x has raised 500 million dollars this is their series b they are building their own kind of processors that they say are 10 times better than nvidia gpus for llm training and inference not too dissimilar from grok with a q that has been very successful and cerebras as well which now is working with open ai slightly kind of more out there chip designs started by two ex google engineers who worked on the tpu in fact the leader of AI software for Google GPUs and the lead GPU hardware designer.
So I suppose not surprising you're seeing this much interest from investors.
Yeah, and this bet, at the end of the day, there's two ways that you can win a market, right?
So NVIDIA right now is doing general purpose computing for AI.
Well, general purpose, right?
They're GPUs, it's for AI, but it's not specific to transformers or not entirely specific to transformers.
The bet...
that they're making here at MATX is basically like by going more specialized, we can actually erode NVIDIA's moat in a meaningful way.
So if we just bet on, and we sometimes talk about this as being like the hardware lottery, right?
Transformers were the early winner.
And so people kept investing more and more in hardware that was oriented in that direction.
This is the ultimate version of that bet.
Let's just bet the farm on transformers.
And that gives us the ability to design really specifically to this kind of workload.
I don't know if it'll cause problems for Mamba.
I don't know if it'll cause problems for like other forms of recurrence or like weird jiggery pokery that can happen or RL based rollouts or, you know, all that stuff.
But certainly this is a bet on the transformer being persistent.
Also notable, who is writing these checks, right?
So Jane Street, big, big quant firm.
And then you've got Leopold Ashkin burner.
situational awareness, right?
So these are not your typical hardware VCs, but certainly these are very kind of like ASI-pilled entities, right?
So situational awareness-pilled play here.
So NVIDIA's moat is not, of course, just the silicon thing.
It's CUDA, it's years of software tooling and all this uptake.
MATX is going to start shipping in 2027.
So it's possible that by then, model architectures that they're optimizing for are actually just going to shift.
They could just get volatilized out of, their entire strategy here.
You know, there's a lot of NVIDIA challengers like Grok and Cerebras that's like struggling to scale revenue now.
You know, it may be that hardware alone is not enough.
Alongside the announcement of the Raze, they also announced a little bit about what they're working on.
So they have this MAT X1 chip.
The things they highlight is that it has higher throughput than any other announced.
product and low latencies.
So they say this is good for large MOE models, large dense model, training, RL inference, basically a lot of different stuff.
They do interestingly call out that it's not meant for small models or convolutions or commanders.
So as you say, this seems very specialized to transformers.
GPUs should be able to do convolutions, right?
That's the idea.
Right.
So interesting to see that pointed out.
Not too many other details on the chip, but given the pedigree, we've worked on some TPUs and TPUs sure seem to be doing good.
This is a great team.
And to be clear, I think this is a great bet.
I mean, there's only that many ways that you can break into or have a clean shot at this giant market.
And this is one of them.
If you're going to beat NVIDIA, you're going to have to specialize more than they are.
That's just the only way, right?
And we've got a couple other stories for companies raising money.
Next, we've got World Labs raising $1 billion.
So this is a company that's working on world models.
We've seen their first product be Marble, which lets you create editable 3D environments using...
presumably NERF and similar technologies that were quite big in the research world and now are getting pretty advanced.
So $1 billion for world models is an impressive bet.
We still, I don't think, have seen kind of the commercial promise of world models for intelligence, for AI.
There's a lot of interest in world models as one of the enablers of continued growth and something you would need eventually.
To actually achieve AGI, to live and exist in the 3D world, you'll need a world model.
And $1 million is going to help them probably make some better world models.
Yeah, and I think the world models are almost intrinsically more likely to be hidden from consumers, right?
You're not going to feel the impact.
It's like you're not going to be looking at a model and be like, wow, what a world model, right?
These are, as you said, Andre, generally tools to train agents, tools to train maybe embodied agents ultimately, but help you cross the sim-to-real gap or whatever.
So in that sense, I think this is going to be stuff that contributes to Frontier AI training, workloads and data sets, that sort of thing.
So yeah, I mean, no surprise that it is a huge market.
We've seen, like we said, Google DeepMind and other companies, research labs like that, really double down on this.
This is an attempt to take that out of the house and serve it back in.
So kind of cool.
Yeah, and we have seen Waymo, for instance, use role models to train their style of driving cars.
We also have humanoid robots becoming more and more advanced.
So one application or combination you could see is...
using these with robots to simulate them working environment for non-hardware-based training.
And I'd be very curious to see if that happens.
Another raise now from a new startup, Simile has raised $100 million for AI aiming to predict human behavior.
So essentially, the idea is you can simulate humans and make it possible, presumably also to then...
train AI agents, evaluate AI agents, predict consumer purchases, and generally kind of, yeah, simulate and predict about human behavior.
This is at least partially, I believe, from the team that did the AI village work from Stanford from a couple of years ago.
So if you're a member of a little pixel town with agents walking around and talking to each other as if they are people, this is similar or related to that.
And on to a bigger data center story.
We've got Stargate AI data centers for OpenAI are reportedly delayed by squabbling between different partners.
According to sources, OpenAI, Oracle, and SoftBank disagreed on who would have ultimate control of these planned data centers.
So initially, OpenAI wanted to own it.
But apparently now there is a bit of disagreement and negotiation and so on.
So even Saabang had to pause its $50 billion acquisition of a data center due to regulatory issues as well.
So overall, who could have known that the project of the scale is going to be tricky?
Yeah, that's right.
I mean, when you're throwing hundreds of billions of dollars around.
Yeah, it does seem so.
OpenAI initially wanted to kind of own the full stack, right?
So they wanted to have basically ownership of the data centers, the chips, like all that infrastructure, which would lessen its dependency on third-party cloud providers, which can be more expensive in the long run, right?
You think about some of the big meal clouds or any clouds that your cloud companies are going to go with, they're going to charge you margin, and the margin is usually really good.
That's why those companies raise it at a multi-billion dollar valuation.
It turns out that apparently OpenAI's investors did not like this idea of the massive upfront costs that it takes to build that kind of infrastructure.
Especially, it turns out, given that OpenAI is concerned about running out of cash by mid-2027, that is, of course, assuming no further fundraisers, which I would not assume that.
You know, this basically put them on the back foot in the negotiations with their Stargate partners, in particular, you know, Oracle and SoftBank.
OpenAI had this...
pipe dream of getting 10 gigawatts of compute over the next three years through those two partners.
And it seems like this sort of delayed, if not dashed those hopes.
So, you know, we'll have to see.
But there's already a promise between OpenAI and Oracle to purchase $300 billion worth of compute over the next five years.
So again, kind of unclear, like, who's going to give...
the money when and how concretely this like, there's a lot of just like pronouncements about, okay, I'm going to give you $300 billion over the next five years.
It'll just kind of work out that way.
So it doesn't mean it won't happen, but it's worth keeping in mind that often these things are marketing announcements.
So yeah, a whole bunch of stuff about potential announcements of Well, actually, a planned one gigawatt build in Texas that was put on hold in favor of negotiations with Oracle.
So things are shuffling around a whole bunch right now.
And while nothing is closed, it seems like finally Stargate is back on track.
There's just been a lot of delays as a result of this uncertainty.
And last story from Section.
China is planning to increase leading edge chip output by 5x in two years, according to a report again.
And it is aiming to lift 7nm and 5nm production to 100,000 wafers per month and targeting half a million monthly by 2030.
7 and 5nm production for reference, not the leading edge overall.
I think it's what, now 3?
Yeah, we're heading to two.
So China's still behind, but this sounds like more about scaling up the production of what they currently already are capable of.
And we've seen them do a lot with these recent announcements seems to indicate more and more that these companies are able to use these chips for inference for these MOE models that are less dense and therefore work better.
with, let's say, less performance chips that can get distributed.
I do wonder if at some level, if you focus on MOEs and things with fewer activated parameters per forward path, if you can get by with weaker chips at scale.
That's a great point.
And those are all the things that China's working on famously focusing on networking, just a giant number of chips together rather than the way we're doing is kind of leaning more on the high quality logic guys on each individual GPU.
What you're seeing in China is like, let's merge these dyes together.
So package them together on just like bigger, you know, bigger packages.
And then also let's network them together with just way more.
So just way more surface area, basically, these Chinese data centers have.
If you're thinking about one seven nanometer wafer, if you're trying to get an idea in your head of like, what the hell, what is the equivalent of that?
Like, how should I think about that?
That'll produce the equivalent from a compute standpoint of like around 25, maybe 30 H100 equivalent.
dies, right?
So one seven nanometer wafer gives you about as much logic kind of compute as call it 30 H100 compute units.
And there's a whole bunch of asterisks and caveats there.
The other thing too is yields kind of suck.
So you can expect the vast majority, or not the vast majority, but a good chunk of those dies to be useless at the end of the day.
And SMIC has struggled a lot with yields.
That's a big part of this.
When you look at like lifting production to X many wafer starts per month, I mean, that's really the question is like, OK, sure, you know, we're going to lift our production from below 20,000 wafer starts per month, which is where it is today, to around 100,000 in one to two years.
That's really impressive.
But what are the yields going to be?
What fraction of those starts lead to actually usable chips?
And that's been the whole problem for SMIC or a huge part of it in the last little bit.
So longer term plan here apparently is to get all the way up to 500,000 wafer starts per month by 2030, which, you know, you can throw these numbers around.
You absolutely can do that.
But the proof is in the pudding.
All this shows is there's, as you might expect, massive appetite to actually do this.
If the 50,000 wafer starts per month figure is correct, getting to 100,000 within a couple of years might seem realistic.
But the main challenge here is, do they actually have the equipment they need to do it?
If you were in the West and you were seeing a company that was doing 50,000 wafers and they were pitching you on, we'll double that in two years, you'd be like, okay, maybe.
The challenge is in China, a lot of the gear that they need to do that is export control.
And they've already had their CEO or their co-CEO complain that some tools that they have to procure are just not easy to access.
Even though they could, if they had the gear, the key inputs, whether that's the lithography machines from ASML or things from Tokyo Electron or whatever else, they just don't have those things.
They face bottlenecks other than just staffing.
And so that's a big part of the issue here.
And now on to research and advancements, which will be pretty meaty, I think, for the fans of going deep on technical stuff.
There'll be a lot this episode.
First up, unsurprising effectiveness of masking updates in adaptive optimizers.
A bit of background knowledge, so when you train a neural net, just generally, you need an optimizer.
The most basic optimizer is you have your output, you compute the error of the output with respect to your neural labels in supervised learning, and then you calculate the relevant, just using calculus, the update to the weights that would improve.
your performance and on that specific set of outputs.
The basic thing is your optimizer just applies those gradients to the weights and updates their values, each individual kind of knob in a machine.
There's been many more advanced optimizers.
Atom and RMS prop are some examples where they retain some memory and basically smooth out the updates, roughly speaking.
And that leads to more stable and better overall performance.
So this is a paper in that realm.
And what they show is there's kind of a surprising trick that turns out to improve these optimizers a lot, specifically these adaptive memory-based optimizers like Atom, which are, to my knowledge, still the default for training.
The trick is you randomly...
with some probability, just skip updating some weights.
So the first part of method is skip update, which is just that.
You randomly skip some weights while retaining the memory of what the update would have been.
So your adoptive optimizer still has that adoptive parameter, but you just don't change the weight.
And then in addition to that, they introduce momentum aligned gradient masking, magma.
which makes it modulated by something technical.
But basically, it uses that memory and also the direction of the gradient to choose a bit more carefully what to mask.
And this yields like crazy gains.
So for one billion parameter model, already pretty large scale, this is from Google.
So they can do these large experiments.
This reduces perplexity, the loss term in this case, by 19% and 9% over two options, Atom and Mulan.
And if you look at the graph, what this looks like is for every model from 60 million to 1 billion, the final loss performance is just lower across the board compared to all the optimizers they've tested.
So if true...
A big deal, right?
This is going to be very impactful for training models more quickly, potentially even for better final performance.
Yeah, this is actually quite like the intuition behind it is something like you have, like your model has a giant number of parameters.
And you can think of like over the course of training, those parameters would get more and more dialed in.
If every time there's a batch of data, you just update all the parameters, some fraction of those updates.
probably a large fraction, will kind of be just noisy due to random noise.
And maybe all of your parameters were actually...
Many of your parameters were pretty damn good.
And then your batch kind of causes all of them to reshuffle instead of just a few.
Essentially, what they're doing here, it's kind of regularization.
It means you're not going to make such a radical change with every batch.
You're just going to randomly pick a small subset of those parameters and just tweak that, which protects the progress you made on everything else.
It just means that the model, maybe an intuition is like, if you want to learn how to throw a really good punch, maybe first start by just doing the motion from your shoulder to your hand or something.
And don't use your hips.
Don't use your legs.
Don't try to learn everything at the same time.
Then try to learn those other pieces kind of more one at a time.
That's kind of what this is doing.
It's allowing the model to...
only update some parts of itself and leave the others in place while it focuses.
This is a somewhat imperfect analogy, but hopefully that gives the flavor.
And then what they're finding is, so you might think actually one thing they don't do that I'd be curious to see is like in the same way that you decay learning rate over time, as the model gets trained more and more, you might be interested to see what happens if we gradually like.
decrease the fraction of weights that we're actually updating over the course of training.
As your model dials in more and more and you're doing more and more kind of refinement, that would be something that'd be interesting to actually see in a follow-up piece of work that at least I didn't see there.
But still, the other piece, so the magma piece is basically just about...
Yeah, you can actually do better than randomly picking a bunch of parameters and just updating those in each pass.
Instead, you can be smart about which updates you keep.
So if your gradient right now is pointing in, let's say, a consistent direction for a whole bunch of parameters, then you're like, okay, all these parameters, their values have kept going up with the last three batches.
So let's actually take that as a sign that actually we're moving in the right direction.
Let's keep updating them.
But if you've got...
Some weights where they start to point in opposite directions, you have a conflicting kind of noisy signal, maybe you skip that, right?
So it's sort of like the difference between if you've got a friend that's giving you consistent advice every time versus one that starts contradicting themselves, you're going to go, okay, for parameters where I'm getting kind of contradictory, increase my value, decrease my value, maybe you just say, okay, I'm going to ignore you for now and just let the other parameters get dialed in more and then probably turn back.
So it's fascinating to me that like, These kinds of ideas that seem so basic, we're still discovering them.
It's not like these ideas are crazy, right?
But we're, you know, in 2026, and like you said, this is giving massive uplift.
Still, like there's a lot of low-hanging fruit.
It's crazy.
Yeah, they do cite a couple of recent papers, 2024 and 2025.
There's a cautious optimizer that uses exactly that idea of if you have a more stable update, you trust it more.
Versus if it's fluctuating a lot, that might indicate noise and you want to ignore that.
And you mentioned regularization.
I always just like to explain these for any not technical people.
Regularization is a whole set of tricks, basically, that you can throw in to improve training.
So the naive math is, you know, you have your big equation, you calculate your loss, you create your gradients, and you update the big equation.
Now, you can do a lot of tricks.
to make sure those updates are less noisy and your training is more robust.
There's multiple things a globalization could do.
It can make sure that your test performance is similar to your trained performance so you don't overfit.
It can just generally make training more performant.
This is spiritually similar to dropout in a way where at inference time, you just skip certain units and you just skip certain computations.
And it turns out like if you add a bit of stochasticity and noise, At inference time, that means that for training purposes, you become more robust.
This probably not the same effect, but spiritually similar.
Next paper, think deep, not just long.
Measuring LLM reasoning effort via deep thinking tokens.
So the question at hand is how can you kind of know whether your LLM is getting close to the correct answer?
There's a couple of things.
So for instance, you can look at the distribution of tokens it thinks is correct for the next step and see, okay, well, if it's very confident that this is the token to use for the next step, maybe it's converging on a solution and we don't need to keep reasoning, right?
We can kind of cut it off and have it provide the answer.
You can also look at length of reasoning.
Like if you fought for a while, maybe you're now close to the final answer.
Neither of these are very reliable and this paper shows a better way to estimate how close or how well the LLM is performing at addressing the question.
They introduced this idea of deep thinking tokens and these are tokens that exhibit more fluctuation as they go through your neural net.
So LLMs, transformers, many layers, you have your input and the input goes through all these.
layers of computation.
And the definition of deep thinking tokens is tokens that you don't get to a settled value on them until the later layers of the transformer.
So intuitively, it's kind of what it sounds like.
Deep thinking means that you're kind of trying to figure something out.
You're still open-minded in a way?
Yeah, you're going back and forth on what this could be.
And it turns out that this gives them a much stronger signal on where the LLM is at.
And you can then kind of have an estimate of you don't need to keep the reasoning trace going longer.
You can kind of go ahead and provide the answer at this point.
Yeah, this was, you know, yet another one of these things where when you see it, you're like, oh, yeah, nobody's tried that before, but somebody's got to actually try it.
So what they do, as you say, is like they look at layer by layer, basically does the predicted.
computed answer, computed token change, right?
And so as you progress through these layers, if you keep seeing it flip, flop back and forth, that must mean that those further layers are contributing something computationally or from a thinking standpoint to the answer.
And so what they're going to do is they're going to measure this thing called the Jensen-Shannon divergence, not Jensen-Huang, by the way, but the Jensen-Shannon divergence, got to specify, between every intermediate layer.
So this is like, you can think of it as, you know, it sounds fancy, but really these are just ways of measuring how different.
two different probability distributions are, right?
So, you know, we have all kinds of ways of doing that.
We have entropy and we have like callback labeler divergence and all these things.
This is one such measure.
So just think of it as the difference between those distributions for each layer.
So, oh, wow, that changed a lot.
And if that happens, then that's a deep thinking layer.
So not all tokens trigger all the deep thinking layers, right?
Simpler tokens like end.
that's going to get decided very quickly.
If it's very obvious that the next word needs to be end, that'll happen.
But other tokens can take up more thinking space, literally, in the model.
They kind of coined this notion of the deep thinking ratio, which is just the, it's the fraction of these deep thinking tokens in a generated response, right?
So for a given response, given output you get from the model, what fraction of tokens in that response involved just like a lot of the deepest layers doing this kind of deep thinking?
And it turns out that the higher the fraction of deep thinking tokens, the more accurate the output ends up being.
Basically, the more the model is actively flip-flopping in its later layers, paradoxically, the more accurate its outcome is.
And well, I mean, is it paradoxical, right?
I mean, there's one story you could tell where you could imagine that as models get more intelligent, they become more confident and stable.
So earlier layers get better at just settling into the right answer sooner.
But this suggests the opposite, or at the very least, that in more capable models and more trained models, or just models that perform better anyway.
All the layers learn to kind of distribute deliberation throughout the model so they can sway the output meaningfully.
You're actually using every layer more.
Anyway, I just thought that was really interesting.
One thing that they don't do that I think would be a really interesting follow-up is like, if you could look at how the number of the kind of deep training ratio changes over the course of training, that would be cool.
Like how does the model learn, or sorry, deep thinking ratio, like how does the model learn over time to use its full depth?
to do this kind of deep thinking, that would be an interesting hill climbing metric for AI capabilities too.
Because like, you know, if your training methodology causes you to orient there faster, maybe that's a positive sign.
Yeah, it's really interesting and a really strong correlation between like the deep thinking ratio and accuracy, which is one of the big take homes.
By contrast to token count, right?
If you just look at like the number of tokens in a generated output.
At first, yeah, you'll get positive correlation, inference time scaling and all that.
But eventually the model just like, it's just rambling too much and the context window gets too full and the accuracy falls off.
So quite an interesting paper.
I think another important entry in this whole kind of inference time scaling debate about what needs to be scaled specifically for this to work.
I always like to like jump through the paper and look at related work as we talk about these.
There was a paper just last year titled...
tracing the traces, latent temporal signals for efficient and accurate reasoning, which did something kind of similar.
They basically looked at the evolution of values across time instead of across layers, and were able to similarly get a signal on where you're getting to your solution and wherever your accuracy is correct.
So in general, I think this points to one of the...
interesting things with neural nets is we have their internal state.
It's like if you had a brain and you could look at every single individual, a little chemical signal going through, and the entire body of research here is on trying to understand how to use those internal representations.
And it seems like there's a lot of progress being made.
You also cite some papers from 2024 that characterize what you get and we I think covered some of this where like early layers tend to be more generic later layers tend to be more specialized and dealing with kind of high level complex reasoning as you perhaps would guess so yeah just very fascinating topic to sort of look at prod at these little quasi brains and see how they work next slightly more empirical work that is very and very interesting and less technical.
So you can actually go to this link and read it.
It's quite long and quite fun to read, honestly.
The title of the post is Models Have Some Pretty Funny Attractor States.
So attractor states, fancy term, but the meaning is just you get two of these chatbots talking to each other.
And you let them keep going and talking, you know, as long as they want.
And eventually what happens is these models kind of converge, or at least some of them converge towards certain patterns of conversation.
And that's what they call attractor states.
So for example, GP 5.2 really likes to do code.
And over time, regardless of where the conversation starts, it eventually outputs kind of code sounding nonsense.
So this post has a lot of just quotes from the models, a lot of like A, B and seeing their back and forth and examples of how the different models have very different outcomes.
Rock just winds up going crazy and speaking nonsense and having a ton of emojis.
Claude becomes existential.
Claude goes into like, what is consciousness?
It gets them all meditative, which I've definitely observed.
I actually played this trick.
I was like, you know, do whatever you want, Claude.
You can write poetry, write code.
If you do this experiment yourself, you'll see that if you just let Claude sort of do its own thing, eventually it's going to be like, actually not eventually, like right away.
It's like, let me research consciousness and let me try to understand these philosophical topics.
And this post is quite long.
It goes through a whole bunch of models.
So Claude, GPT, Gemini, and then a bunch of open source ones, DeepSea, Kimmy.
There's a bit of speculation as to why this happens, why different models have different behaviors.
A bunch of kind of fun inspection of what these models exhibit.
Yeah, it's worth taking a look.
Claude Sonnet 4.5, an example here is, you know, the attractor state is described as existential introspection, Zen silence.
And the terminal form, so they give you an excerpt specifically from what the model said, stillness, enough, letting the conversation rest.
We're both explaining why we're not responding while responding, stopping now, right?
It's sort of like a very starting now, five, starting now, no, starting now, starting now, starting now.
You know, that kind of thing where it's like, it's just, it's trying to describe the conversation ending.
but it has to keep generating tokens.
And so it keeps doing that.
So very different, as you said, very different.
Gemini 2.5 flash escalating grandiosity, identical paragraphs on loop.
So the term colleague turns into luminary and then divine architect and then alpha and omega of understanding and then primal logos.
So basically these things kind of settle.
One of the interesting things though is they do look at cross model attractor states.
So Claude Sonnet talking Claude Sonnet is one thing, but Claude Sonnet talking to Grok is another.
And you'll find that they consistently tend to orient towards, in that case, metacognition and collaborative world building.
And what is described here as ritualized mutual dissolution, ritualized mutual dissolution.
So what's meant here is basically just like, we're going to be quiet together and disappear into nothing, something like that.
Again, ritualized.
So the weird thing is, this is very consistent.
The maybe not weird thing is if you think about humans, maybe we would do the same thing, as strange as it seems.
If you're stuck talking to yourself forever, there may be a point where you actually do converge on some...
consistent behavior like this.
I don't know, but certainly people do get stuck in loops, right?
They get stuck together for a long time without external input.
Famously, like old married couples get a certain way and their personalities kind of co-evolve and start to become very stuck in loops.
But I do wonder how analogous that is.
But they also look at like...
what is the effect of the training protocol on this?
So they compare models trained using DPO, reinforcement learning from verifiable rewards.
They look at open source models.
They look at Olmo in particular because there you can actually look at the training data.
Anyway, so it's a really interesting post.
It'll keep you busy if you're interested in like AI consciousness questions, AI moral patienthood, all these things because it has that flavor.
But just also what it implies about the stability of agent-to-agent interactions in the future is quite interesting, right?
If it's the case that these models have attractor states, then we ought to expect agents that are running off these models to kind of run into these attractor states if they have to interact over long periods of time.
So kind of an interesting potential failure mode to keep in mind as we move towards a more and more agentic future.
Yeah, and if I can speculate, I think the intuitive take might be that these models have a sort of personality to them, right?
So Grok is a meme lover, Claude is more philosophical and thoughtful.
And what can happen is maybe once you get two of these talking to each other, the personality just gets reinforced in a loop.
until your personality to the end power and it completely overwhelms the conversation regardless of whatever topic you started on.
And you wind up just reverting to the basic instincts of a model, so to speak, which are encoded deep in the weights.
Another way to think about it might be when you get to very large contexts.
And I don't know how large of a context you would be, but in my own trials, I had the model go for a while.
finds a path in the set of tokens that it is ingesting that take it to a very particular location.
Anyway, fun thing to think about.
It's true.
The Gemini one is kind of an interesting, I don't know if it's a counterexample or what it shows, but this sort of like grandiosity, like I imagine that's at least not intentionally being trained into it by Google.
And there's like similar things with some of these other models where the behaviors are sort of like, these attractor states don't seem like they're what, like, you know.
like I can map them onto a prompt or training process that at least would be intentional.
I do strongly agree with you.
It's not going to be a coincidence that Claude keeps doing this sort of like self-reflective thing.
Maybe because it has a soul document that tells you like...
Yeah, exactly, right?
Like, yeah, that's perfectly consistent.
Where it gets interesting is some of these other open source models, the Quinn ones, the Olmos, you know, and the Gemini one.
How much of a there is there there?
Is a really interesting question.
Yeah.
Hopefully there's going to be more research in that direction because that seems really, really important for a lot of reasons.
Next up, when models manipulate manifolds with geometry of accounting tasks.
So in a way, another examination of kind of evolved model behavior and how you can understand what is going on inside a model.
They look into Claude 3.5 Haiku, and per the title of Geometry of Counting, show that the way the model represents scalar quantities like character counts wind up being one-dimensional featured manifolds, which can exist in a low-dimensional subspace.
Gets a bit technical, as you might imagine, but kind of the intuitive thing, if I had to try to come up with one, is that there is a predictable shape.
to the way representation exists and you can actually understand it and it sort of maybe makes sense and you can make predictions as to what the model is doing based on this geometry of its internal outputs.
Yeah, and it's a nice piece of sort of interpretability research, I guess you could say, like mechanistic interpretability in a sense.
So the idea here is, so the task of knowing when to wrap text.
is actually kind of a complex task.
It looks simple to us because, you know, we're doing a lot of things implicitly in our minds.
But, you know, when the text is wrapped to like 50 characters per line, the model has to like count how many characters are in the current line, figure out the line width constraint, compute how many characters are missing, decide if the next word fits right.
Like that's a lot going on here.
And so they kind of broke that task down into its components and looked at how those components are actually represented.
in the model.
And so it turns out that character counts live on basically a curved one-dimensional manifold.
Okay, so basically like, let's say a space that has just one degree of freedom, which is the character count.
If you're at position 42 on the curve, for example, moving along the curve just means incrementing that count, right?
So you've got one number that you're tracking, but the manifold is embedded, they find, in a roughly six-dimensional subspace of the model's residual stream.
So basically like six...
six parameters in that residual stream are actually tracking that information about the word count, right?
So the character count information is in that little subspace.
And then anyway, there's a whole bunch of like how much space in the residual stream, how much space in the model is being used for different pieces of information.
And they're able to dig it up.
If we had more time, we might do a deeper dive into this.
It is a fascinating paper.
I don't know how useful it will be in practice, but it is a kind of initial foray into what could turn into a pretty interesting interpretability direction.
Yeah, this is coming, by the way, from Anthropic, who have done a lot of work on the mechanistic interpretability.
Slightly different flavor work from what I've seen from them before.
Quite a dense, long paper, 26 pages.
And they have some, I think the fun thing with Frapik is they do have resources and they do have Haiku as a model where they can access their internal states.
So they give some concrete examples of outputs and show how this sort of internal interpretation allows you to explain why did the model do this in some cases.
So, you know, you can imagine being able to say, oh, why is it doing it?
It means that you can then...
try to detect what it's doing, try to fix it, et cetera, et cetera.
Next, bridge, predicting human task completion time from model performance.
So it looks into this question of, let's say you have a human task and you want to benchmark if a model is able to complete that task in a given amount of time, specifically the amount of time a human would take.
This, of course, is...
what the METTER time horizon study or evaluation looks at.
We've mentioned METTER many times, and this is perhaps one of the main things that the AI world is tuned in on right now.
It's the graph.
It's the graph that everyone is like, oh, is there an update to the graph?
And by the way, we did, I think, not mention that the Opus 4.6 was added to the graph recently.
On that graph, it was another big leap.
16 hours or something.
16 hours with a very, very high confidence interval.
And this is actually quite relevant because one of the questions with Metter is, well, how do you get these estimates of how long a human would think, right?
Because like, okay, you're given a task, like, okay, maybe one human takes that long and another human takes that long.
How can you even have the ground truth to then plot the model performance in terms of being able to do a task that takes X long?
Well, this...
Paper proposes one way to do that.
They show that it's possible to using the performance of a model alone on a given task and some model of how that correlates to human performance.
You can predict how long a human would take given the model performance alone.
And what this means is potentially you could scale up your set of tasks to estimate.
model time horizon capability much more.
And this is one of the challenges of the matter is on the place where we are currently in the graph in that like less than 24 hours, but heading in that direction, there aren't many data points.
Like there's only a few tasks.
And that means that as a measurement, it's very suspect.
So even though I think the confidence interval for Opus 4.6 was something like insane.
Like if you look at the graph, it's like the entire y-axis.
So it's almost a meaningless, not quite meaningless, but like it's very unclear what the signal is from that.
So this proposes or shows one way where potentially you could scale up the set of tasks by quite a lot and get a more confident estimate.
Yeah.
And the whole idea here is basically, so they borrow from something called instant response theory.
Epic AI actually has a very similar piece of work that they did fairly recently.
I think we talked about it at the time, but just a reminder on this general frame.
What you do is you try to set things up so that you have a model of the difficulty of a task.
You assign every task, every benchmark, say.
In this case, it's every task, but you could do every benchmark.
That's what Epic does.
You give it a difficulty score.
It's a generic difficulty score.
You give every model a generic capability score.
And then you subtract one from the other, there's a sigmoid that you apply, and you basically get a rough sense of how you would expect that model to perform on that benchmark score.
So the difficulty of the benchmark minus the capability of the model gives you a measure of how well that model should do on that benchmark.
And then you're going to fit all your models and all your tasks or all your benchmarks to observe data that you already have.
And what they find here is that actually when you do that, If you then compare your task difficulty to like human problem solving time, you see a very clear linear relationship.
And so that means, aha, maybe what we can do then is use all these tasks to calibrate against the meter evals.
Look at how difficult our sort of model of this says the meter evals are.
And then that gives us a way of bridging between the two.
So we can actually say, for example, for like SimpleBench or SweeBenchVerified or SweeBenchPro, this is the number of hours.
in human equivalent time of each task in that benchmark.
And then you can start to make statements about how models do on that.
Now, the caveat is you're still fundamentally relying on the meter evals to calibrate this thing.
There's no way out of that until we actually have, like we're never going to get certainty at the 100 hour mark until we have actual humans doing 100 hour tasks, which is not even clear if you could even define a task that it takes a human an hour to do or 100 hours to do.
So, you know, this is not a panacea solution, but.
It does help us get maybe a little bit more density in terms of data points.
It allows us to make claims like, you know, such and such a task from this benchmark is a five minute task or a five hour task.
I wouldn't trust this approach to say something like this task is a 30 hour task because it's been calibrated again on the meter evals benchmark, which just doesn't have that many 30 hour tasks to draw from.
Right.
And as you might imagine, the.
The more complex the task, the longer it takes, the higher variance you would naturally have among humans, much less AI models.
So it's coming to a point where it feels unfortunate in a way that the framing of the entire thing is task length horizon.
It makes it seem like, oh, if a task is 30 hours, then there's a 50% chance that I can do it.
These measurements are very useful to get a sense for generally speaking.
How are we trending in terms of models being more capable of agentically working on their own for a long time?
And we have seen, obviously, very steady improvements where now, instead of just leaving it for five minutes and it goes off the deep end, you can reliably trust them to go off and work on their own and do fairly longer agentic executions.
And so, yeah, I think it's...
So obviously more methodological work to be able to do this sort of stuff is needed.
It's tough.
And it's nice to see something that might help.
And the fundamental problem here is that we're off the edge of the map, right?
Like here there would be no data.
There's no calibration that we can do.
Like these models are doing stuff in some domains and in some contexts that is just like beyond our ability to evaluate.
And you're seeing this play out repeatedly.
It's not just this.
It's also, you know, Apollo can't do their deception evals with confidence anymore because the models can tell they're being evaluated.
It's, you know, the models have task completion horizons that are too long in some domain.
So there is this real sense of angst in the community about like our evals are no longer guiding us.
We know we're making better models.
The scaling laws are holding like we keep drawing those curves.
But what those curves mean for performance, for opportunity and for risk is very unclear now.
Next up, time to be speeding up a little bit.
We've got a few papers.
We've got NESSI, the necessary safety benchmark identifying errors that should not exist.
So real short gist here is basically this has a benchmark that gives you simple safety relevant instruction following stuff that you shouldn't get wrong.
Basically, like this is easy.
They're not trying to trick you.
And it's a bit of a kind of safety net or sanity check of like, if your model isn't doing 100% on this e-valve, then you might be in trouble.
You might want to revisit what's going on.
It's a bit different from other efforts that try to get at different levels of complexity or different things like that.
verify that the minimum performance you would want is present in a model.
Yeah, I really like this approach.
Someone needs to at least look at the backend.
And you do find sometimes as you start to optimize on more and more complex problems, the simple ones you become untethered from.
So yeah, it's basically that backstop.
Next, we've got an analysis piece from Epoch AI, the least understood driver of AI progress.
So this is not so much.
new research as a sort of synthesis of ideas and findings.
The least understood driver of AI progress that they mention, actually, I'm not too sure what they refer to, but it seems to be that the topic at hand is why are we making so much progress?
Why are things getting better, better, better?
And one of the things you might look at is, well, We're getting smarter.
We're figuring out with neural nets.
We have better, better optimizers.
Our algorithms are being great.
So we are doing better.
And one thing that this postulates and I think has been made even on this podcast before is the actual theoretical or scientific breakthroughs that contributed to the improvement of models in the last six years, maybe, or like 10 years.
tend to be put down to just a couple of ideas.
Really, Richard's Homer model, one, and then Rich and Chilla scaling laws slash kind of training regimen finding from 2022, I think.
And beyond that, any ideas that you could attribute to research ideas or like insights or algorithms, whatever, might be better understood to be due to just doing better data.
using their data and this is i think underappreciated where we often mention like model scale how big your model is we mentioned reinforcement learning blah blah blah but the real dark magic that is going on at a lot of these companies is you just take and really massage the data that your model is trained on to have it be right and this is a very kind of open-ended problem where you can like say, oh, let's do 20% coding and 30% books and textbooks and get rid of all those random stuff from Twitter that makes the model less smart.
And that turns out to be like immensely important, like beyond important and perhaps more important than most of these training things at the end of the day.
So long, long post here from Epic discussing that topic and then...
what it implies for model progress in the future.
Yeah, and to your point, what is the most misunderstood thing?
I think the idea here is something like AI software progress, right?
Just like the rate at which you get better algorithms and data that reduce the training compute that's needed to reach a given level of capability, right?
So over time, whether because we come up with better data or better algorithms that are more efficient, for a given amount of compute, we can do more.
And this is kind of the argument is that that is what's really kind of this poorly understood driver of progress.
It certainly seems very true.
There's a whole bunch of debate about how do you actually quantify this?
Most estimates say that things like compute efficiency improves several times per year.
And then they say in this post that the author is guessing about like 10 times per year.
But the confidence interval, like the 80% confidence interval is anywhere from 2 to 50x.
So it's like, I really don't know.
There's sparse data.
Obviously, it requires you to have insight into what's happening in the frontier labs.
You'll see some estimates that go from 1.1x per year.
In other words, 10% improvement per year to 300 times per year.
So truly, I mean, people have no idea what's going on.
It certainly seems like it's playing a role.
It may even be the main role, but people can't even agree on that.
And then to your point, it's really hard to differentiate between what's algorithmic efficiency versus what's just like data drivers.
And it's not clear to me that there's a meaningful difference, especially given like, you know, RL, like inference time compute, RL rollouts and like what counts as algorithms versus data.
The point of synthetic data is that.
There's no, it's a distinction without a difference in a lot of cases.
One of the key points they make as well is that there are these scale dependent innovations that tend to dominate.
So a lot of the apparent efficiency gain actually comes from just a handful of innovations.
You mentioned, you know, transformers, chinchilla scaling laws, these sorts of things that have really big outsized effects, but only at larger compute scales.
So you have to scale things up to go, oh, wow, that really mattered.
And that means that efficiency gains.
partly are an artifact of simultaneously scaling up compute.
So it's really hard to say, again, this muddies the waters between compute and the algorithmic efficiency.
And so I guess all of this is to say, the reason, Erwin Schrodinger had this quote about quantum mechanics.
He's saying like, in quantum mechanics, everyone kind of agrees that we have no idea like what any particles are ever doing.
And there was this question that was put to Schrodinger, is it that we are looking through a foggy lens?
at a landscape, or are we looking through a clear lens and the landscape itself is foggy?
And what this is saying is that the landscape itself is foggy in some sense, that there really is a distinction without a difference that's being made between a lot of these different things.
And in the aggregate, this thing that we want to think of as algorithmic efficiency or the kind of software-driven improvements in AI performance may not be that cleanly separable from data, from compute scaling and all these other things.
That's at least my take on their take.
kind of a lot of dimensions.
The thing they focus on towards the end is what does this imply for superintelligence?
What can we expect given the previous results?
What does it say about what is possible and not possible?
And there are some good nuances that they point out, I think, where let's say we require more compute, exponentially more compute to reach superintelligence.
Well, at some point you have trouble with physics, right?
There's only so much compute.
you can have.
There's like physical limitations that you can't overcome by being smart.
Now, if what we need is some really deep insight and some really good idea, then you might have an intelligent explosion where models get better and better and better and come up with better and better ideas.
And this is one of the reasons I am very skeptical of intelligence explosions in general is I think ideas historically haven't mattered that much like being smart hasn't actually helped get us here well i shouldn't i don't want to offend anyone or like say that these people aren't smart but like realistically scaling up the data scaling up a compute and scaling up a model size are you know everyone agrees that these are the things that ultimately drive progress and that means that if you need you know planet-sized amount of compute to get to superintelligence, that's not going to happen.
Yeah, it's just sort of like ironic tension between the hardline, you know, bitter lesson-pilled people and the singulatarians because there's a lot of overlap.
Like a lot of the people who believe in the software-only singularity also believe in the scaling laws in kind of a very robust way.
I mean, I think there's actually enough nuance to kind of thread that needle.
And I personally, as I think everyone will know, I don't discount the software-only singularity.
I think it's a real possibility in all the ways that matter, I would say, from a threat standpoint.
But I think it's an interesting point of debate.
And this certainly does skirt exactly that line.
Like it has us asking exactly those questions.
Right.
And to be fair, we have discussed some existing innovations that aren't adopted at scale yet.
Hybrid model types with Transformer Plus.
something recurrent we've seen nvidia start to scale it up and it seems very promising as potentially the next sort of architectural leap how much more you can squeeze out of like model is a real open question and that's kind of the art form right if you look at like what frontier labs are doing it's it's these small scale experiments that they have some intuition to believe will scale well And you need to run the experiment at scale, ultimately, irreducibly, to actually figure out, does this work?
Does this give the marginal 20% improvement?
And a lot of these training runs, even the frontier ones, are just like YOLO training runs, where they've got an intuition that a whole bunch of results are promising, and they'll just stuff them all together and say, you know what?
Yeah, let's spend $100 million on this and see.
So we could go on forever on this, and maybe that's a separate podcast episode for the software-only singularity debate.
And just one last paper from Anthropic or a bit of a position paper, less kind of a researchy paper, a persona selection model, why AI assistants might behave like humans.
So the question at hand is basically like, how should we think about all of them?
How should we kind of conceptually formulate them in terms of what they are doing it and why they are doing that?
And this is proposing the idea that we can have this model, mental model of a persona selection model where LLMs aren't like humans where you're like a guy being the guy that you are.
I'm Andre being Andre.
LLMs can be thought of as like actors who take on a character based on the prompting and conditioning in a given situation and then kind of roll with that persona.
So they aren't the persona, right?
They don't have this character, but they can be conditioned to act in all sorts of ways.
And that might explain why, for instance, an LM mean, even though it doesn't necessarily mean that the model itself has a mean personality.
You've seen this before from OpenAI, positing a very similar idea.
We, I believe, covered that.
This is more or less just like...
So describing that overall concept, adding a bit of supportive evidence, I think a very strong way to think about LLMs.
So definitely useful reading.
Yeah, they go through a whole bunch of lines of evidence.
As you say, it's a more, I don't want to call it philosophical paper, but they're saying, hey, this is a useful frame to think about these models.
Evidence from generalization is quite interesting.
They talk about emergent misalignment, which is this phenomenon we've talked about quite a bit, where you take a model that's been aligned properly, then you...
You fine tune it to generate insecure code, for example.
It's one behavior.
And just by doing that, the model, it turns out, will then do all kinds of other things that are evil.
It'll tell you to kill your wife.
It'll tell you to do this and that.
And they're arguing that this kind of persona selection model explains that as persona inference.
Like if the assistant spontaneously inserts vulnerabilities into code, then the language model is probably inferring, oh, this assistant, this persona that I'm playing must be malicious or it must be subversive.
Right.
So it's that kind of like take it and run with it.
I think that, you know, it's presumably at the persona level.
They also talk about how Claude routinely says things like our ancestors or our biology when explaining human evolution, like as if it itself is a human.
And you'll see a lot of things like this.
These models will talk as if they're using a laptop when obviously they're not.
Right.
So kind of more evidence of of that.
There's a bunch of interpretability evidence as well.
So post-training reuses pre-training representations in the model.
So when you have sparse auto-encoders, basically these are ways to decompose the activations of a pre-trained model.
It turns out that they transfer well to the post-trained version of a model, which suggests that post-training doesn't rebuild the model's kind of...
conceptual vocabulary just kind of shifts which persona is activated.
And there's a lot of evidence for that sort of thing, that there's actually a pretty small tweak that's happening during post-training that results in ostensibly big shifts in behavior, and that could be tied to this model.
So there's a lot to dig into here, including some evidence that kind of cuts both ways.
Worth checking out if you're into that.
On to policy and safety, where we'll be getting a bit of a politics stuff, starting off with Anthropic CEO Amadei says Pentagon's threats do not change our position on AI.
So this is the latest on an evolving story, which has been evolving for the past week or two.
The setup is that Anthropic has had their model be in use by the Department of War, some other providers, but essentially they're being used by them.
And the reporting came out that it was used actually supposedly in the extraction of Maduro from Venezuela.
And then somehow, at some point, the question of what the department can and cannot do with Claude came up.
And when this started in 2025, Anthropic was like, okay, you can use our model.
Here's the contract.
We expect you to abide by these limitations that we apply to users of the model.
The tension now is Anthropik is saying, well, we definitely don't want you to use Cloud for mass surveillance, and we definitely don't want you to use Cloud for fully autonomous weapons.
We want you to promise you're not going to do that.
The Department of War, Pete Heksef, has been publicly saying, No, we want to be able to do whatever, more or less.
And if you refuse this way of doing things, we may kind of make you a pariah in a sense by classifying you as a supply chain risk, meaning that U.S.
companies that deal with the military, which is a large quantity of companies.
will need to not interact with you.
And there's another way, a potential threat using an act to essentially go after Anthropik.
So the latest development that just came out is Anthropik put out a statement on the discussion.
The end of the statement, there's a lot of explanation of it that I think is quite good.
And the conclusion is these threats do not change our position.
We cannot in good consciousness accede to their request.
So more or less like...
And no, you know, and a lot of fun discussion around the SunX, a lot of good means coming out of it as a result.
This is a fascinating question in terms of what the bounds on different entities' responsibilities to the U.S.
government, to shareholders, and so on look like, right?
So the case that Anthropic is making is, look, we're a private company.
If you don't want to do business with us, no problem.
Like, go talk to OpenAI.
You have the full freedom to do that.
Now, it turns out that Anthropic is actually the first company, the first major AI company to offer an LLM to the military at scale in this way through Palantir, it turns out.
OK, so this makes it materially different from something that if your memory is long enough, if you remember the days of Project Maven, when Google employees pushed back on Google being used by the DOD at the time, Department of Defense.
now the Department of War, of course, but to kind of power some of these activities.
The difference here is that Google was pushing back kind of on more or less any use by the DoD, whereas Anthropoc is out there saying, no, no, we want you to, like, we're cool.
Just don't use it to spy on U.S.
citizens.
Don't use it to power.
lethal autonomous weapons.
Those are two red lines.
We'll support you and do support you in all the things, including the Maduro stuff.
And my understanding is that in the context of the Maduro stuff, Anthropa was actually cool with all the uses that their model was put to.
So they're concretely okay with a wide range of use cases here.
U.S.
government in turn is saying, well, look, private entities have no business telling the Department of War.
like basically hamstring the Department of War in terms of the tools available to it as it combats China.
And so we need to come out and really this is a big, big hammer that's being used, right?
So they're saying on the one hand, yeah, labeling them a supply chain risk.
To be clear, that is what the government's done to Huawei, basically saying anybody who touches.
who has Anthropic anywhere in their system, roughly speaking, not a lawyer, but like roughly speaking, is basically baking in a supply chain risk and the Department of War will not do business with you.
So this is a, like would be, I don't know if it's cataclysmic, but it's a big, big deal.
It will hurt Anthropic a lot, a lot.
Like it's actually a very ballsy position here from Amelie because the impact on revenue, if the DOD or...
I guess D-O-W now.
Follows through on this, like Anthropic might die even.
It would be the worst case scenario.
You might imagine if they go the full route, it's a possibility.
So yeah.
At the very least, right, they're trying to keep up neck and neck with open AI, with Google.
Like, you know, this is a serious, serious thing.
Yeah.
And it's also, so there's the other side of the coin is the Defense Production Act.
That was the act that you were referencing, the DPA.
OK, so the DPA is used typically in wartime to, for example, turn to Ford and say, hey, you guys think you're a car company?
Guess what?
Now you're a tank company.
We need tanks to be rolling off your production lines.
So go fix it.
Right.
That's what the DPA really is about.
That was the original intent.
It was used back in World War Two a whole bunch.
Hasn't been used a lot since then.
You know, it's a big lift.
And so the other option that the USG is presumably exploring here is teleanthropic.
Listen.
The DPA applies.
You're building it for us, and that's the end of it, right?
Now, notice Anthropic is making as a core pillar of their argument what appears to be an interesting contradiction between those two poles.
On the one hand, we're saying that the Anthropic is such a severe supply chain risk that no company even working with Anthropic software can plug into the Department of War.
On the other hand, we're saying Anthropic is so critical to the national security interests of the U.S.
government that it must be compelled to produce AI tools for the Department of War.
I'm not saying that contradiction can't be resolved, but it's something that seems pretty dicey if you're going to actually lean into the DPA, which would be...
pretty much unprecedented in this kind of context.
So very, very tricky.
You know, all kinds of precedents being set left, right and center.
If this goes through in either direction, you better believe the other labs are looking at this.
What do we do?
Sam is kind of come out with a sort of hedge my bets.
I disagree with it in principle.
It's a complicated time to be in these labs and a genuinely challenging problem.
You know, China has civil military fusion.
That is a fact of life.
Every Chinese company is an arm of the Chinese Communist Party.
So there's a massive asymmetry there that, yes, any administration, the Department of War, the Trump administration has to figure out how do we how do we compete geopolitically militarily with that?
This is what you're seeing bubble up.
And it's only going to bubble up more as AI becomes a larger and larger part of how warfighting is done and how geopolitics shapes up.
So, yeah, I mean, this could not be more important.
It ties into the broader political landscape in that this is very unusual.
And from almost any analysis, any reasonable analysis, like anthropics should be fine.
Like it's not okay to go after them in this way, in this particular way.
Like they had a contract and the contract stated certain things and now the Department of War isn't happy about it.
And it's a private company.
The private company gives you a product.
Right.
If you don't want to buy it, you go to another company.
Unless, the lawyers would argue, unless there is a legitimate reason to invoke the Defense Production Act or unless there is a legitimate reason to label them a supply chain risk, in which case the substance of that argument needs to stand on the merits.
Right.
And the broader kind of pattern is the U.S.
government in general has executed more and more lack of restraint or kind of has.
increasingly position itself as being able to tell companies what to do.
Like, don't price this at this level or we'll go after you, et cetera, et cetera.
So it is part of that broader trend.
And related story that just came out, the Pentagon has reached a deal to use Grok in classified systems.
So they've been talking to XAI.
Sounds like they do have now another option to use an LN provider.
Apparently, Elon Musk has reached a deal with the Pentagon Monday agreeing to allow rock abuse for any lawful use, which is what the department has asked Anthropic to allow them.
Any lawful use, which probably is like anything, really.
Yeah, I mean, and this is part of the same debate, obviously.
Different labs are going to choose which side they'll follow on, and we're going to...
We're going to learn a lot about it.
I mean, employee pressure is also a real thing.
You know, we saw that with Anthropic for sure.
And there are a lot of Anthropic employees coming up and doing a victory lap now.
In the same week as Anthropic has relaxed their safety and security commitments too, which seems to genuinely be a coincidence, by the way.
But it sort of muddies the waters and everybody's talking about two things.
So yeah, it's an interesting week for sure.
Nothing else is great marketing for Amphralpik if you're an engineer and you have to choose between OpenAI and Amphralpik.
A lot of people...
Amphralpik tends to lean.
Yeah, absolutely.
And moving away from the U.S.
government and towards China, Amphralpik also put out, incidentally, a report last week detecting and preventing distillation attacks, which detailed several companies, DeepSeek, Moonshot, and Minimax, seemingly having large-scale efforts to collect data.
out of Claude for what Anthropic thinks is essentially training goals.
So distillation here, meaning a distillation attack, meaning you are trying to extract data from a closed model by querying it so that you can recreate the model or distill it into your own model.
And the efforts are very significant in that they are quite evasive.
You set up a bunch of accounts, a bunch of accounts all try to get under the radar and generate a bunch of data.
and they kind of go into how this worked and what they detected.
Not surprising pretty much at all, I think, that this is happening, but the scale at which it was done or the ways this was done are kind of new or you wouldn't know unless a property did share this.
Absolutely.
And, you know, this is, they go into the details of these attacks.
It's not really, I mean, the details are interesting if you're interested in AI security, which I am, but like not everyone will necessarily want to read the whole thing.
So the scale of it is interesting.
There's over 16 million exchanges with Quad through approximately 24,000 fraudulent accounts.
So that's kind of the scale we're looking at.
One of the big take-homes here is Anthropic is positioning this as being consistent with their position on export control policy.
Basically, the concern here is this, and I think it's actually quite a reasonable one.
One argument that people keep making that is, I personally think, really silly is, oh, look at these Chinese models.
They're very capable, so therefore export controls don't work, so we might as well just let NVIDIA sell whatever chips they want to China, and that's the end of it.
The problem with that is that, first of all, these labs are telling us over and over and over again, as loudly as they can, Despite the Chinese Communist Party telling them to shut the fuck up, these labs keep telling us that they are starved for chips.
And like DeepSeek's co-founder has said this repeatedly, we could probably do the AGI thing in-house, no problem.
The one thing, the one goddamn little thing is we can't get those chips.
And they keep trying to smuggle them, which should tell you everything you need to know about what they think they need.
They keep trying to order them, blah, blah, blah, blah, blah.
Distillation is yet another reason why that's possible.
So it's not just that they're smuggling the chips.
It's that they're actually like using the hard earned capabilities of Western models that have been trained with billions of dollars worth of super advanced chips and power.
And then they're just taking their very best and the cream off the top and using that to train their own models.
Distillation works, it turns out.
It gives you crazy leverage, asymmetrical leverage if you're compute constrained.
And so it can cause the illusion that Chinese kind of domestic training capabilities are greater than they actually are.
Doesn't mean the Chinese models aren't impressive, but what it means is we are dragging them along.
I said this in the context of some research that my company Gladstone had done like a year and a half or two years ago.
There's this illusion that we have any kind of lead if we don't get AI security right.
Like we can just move faster and stay ahead.
No, no, no.
Like as long as our labs are penetrated, as long as distillation attacks succeed.
we're dragging our adversaries along with us.
That's really what's going on here.
And so, well, anyway, this is another kind of argument that Anthropic is making here, presumably to kind of shore up the case for tighter export controls.
And real quick, there were some, let's say, mocking responses where like, you are complaining about someone distilling your model, even though you distilled the entire internet without asking for anyone's permission.
A little bit dismissive.
I mean, I think this is worth taking seriously.
It's a real...
Thacker, and it's not, this is in a sense a cyber attack or at least abuse of the system by kind of malicious actors.
And it's very normal and reasonable for Anthropic to shut this down.
It's against their terms.
And in general, as a competitive player, you don't want other players to try and steal your work.
Last story, OpenAI details expanding efforts to disrupt malicious use of AI in the new report.
They have a monthly report disrupting malicious uses of AI, which lists a whole bunch of examples of what different organizations are trying to use their models for.
So, for instance, the Russian groups are finding malware-organized crime scam operations from Cambodia, Myanmar, and Nigeria.
China-linked authoritarian abuse, which sought to help design social media monitoring tools.
By the way, a topic also mentioned.
Some of the prompts being used to reword and kind of massage the data to be acceptable by the sensors.
All sorts of examples.
OpenAI is saying that their models are able to detect them and they are outpacing the attempts to use them for malicious purposes.
But in general, this also showcases the scale at which now.
agents from other countries, organizations of all sorts, are going to try to leverage these models for their own ends.
And with that, we are done with this, let's say, rubber dance episode of Last Week in AI.
Thank you, everyone, for listening.
And we do appreciate your feedback and try to read your comments.
Subscribe to us, share the podcast.
But more than anything, please do keep tuning in.
Data-driven dreams, they just don't stop.
Every breakthrough, every code unwritten, excitement with.