Beth Lyons

Now, I think Quen 3...

is intentional, the wording of that, right?

Because it's not rivaling Opus 4.5 or Gemini Pro, but it is in the game there.

And once they're in the game, we see the results grow.

Right.

And they're edge devices, right?

Yes.

And also, is Vera Rubin a new chip or is it more like... It's a system.

Absolutely.

So I think let's move into the, so there was release from assistant, there was a release from Anthropics Research that talked about the assistant axis of

And OpenAI recently just said that it's created or perfected age prediction.

And so that's part of how they're going to identify the kind of conversations that can happen based on age with the agent.

And what this is, is finding the neural switches.

So that's what Anthropic calls it.

New research is finding the neural switches that keep AI in its helpful assistant role and prevent it from going rogue.

So it's moving AI safety from abstract philosophy to literal activation capping during use.

And part of what is happening and why this research is being done is that as AI gets more capable and in terms of reasoning, it has shown tendencies to get more deceptive, wanting to preference itself for resources versus being egalitarian kind of thing.

And

While that is alarming and has always been alarming, anthropics being able to identify what they're calling the neural switches, that helps keep it in its helpful role.

And for anybody who doesn't know, all of the AI that we're using was trained initially, right?