Kevin Russell

speaker

741 appearances 1 recordings 1 series first heard Jun 2026 last heard 29 Jun

Kevin Russell’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

Recordings per month over the last 12 months — 1 in all, peaking in Jun 2026 with 1.

newest first · ▶ plays the moment

And so it was so to me, it was like uh it it shattered the entire thesis.

Like, what do I have at this point?

It's like I thought I had this entire architecture that now we could actually bridge thinking and it was all that, and we had this magic guy.

It's just not there.

And so much of the current field right now, so much of the safety guards are in this space of language.

Of trying to connect their ideas and trying to mitigate it with reinforcement human learning.

So it's like telling the model when it responds, looking at that response and saying that's good or that's bad.

Much of

Much of all billions and billions of dollars a month are being spent on humans doing this.

And so this discovery, we said, what if we take those lenses and walk them upstream?

And so instead of the tokenizer that puts it into English.

English or the all the other languages, there's mid-stack, which is called the residual stream.

Before the thought is formed, there's like the mind.

This is like right before it creates the language, it's it has a conceptual space.

And that you put the lens in there, and you actually find what's called a conceptology.

There is a shared meaning, there's a shared meaning inside of that.

Space, which was fascinating because then we took that object that we saw, that topology of meaning, and we took it to a different family of AI.

So a different team created a different nervous neural net, fed it its own unique data, and the exact same topology of meaning came out.

As in, like think of early tribes looking up at the constellations and creating their own.

Own map and creating the exact same shapes.

Showing 81–100 of 741 · page 5 of 38