David Cox

speaker

634 appearances 1 recordings 1 series first heard Jan 2026 last heard 13 Jan

David Cox’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 Jan 2026 with 1.

Appearances

newest first · ▶ plays the moment

The Neuron: AI Explained · Why IBM Wants AI to Be Boring · 13 Jan 2026 podcast

And actually next, you know, the next however many tokens I want those weights.

So you can kind of just pop them on and off, which lets you do kind of like fixed functions.

So like now you're the best hallucination detector in the world.

Like tell me if there are any hallucinations in there and it'll like return structured form.

So it gives you this kind of dynamic ability with a small model to,

sort of lean its capacity.

And we can also do the same, you know, you can imagine doing the same thing with personas, like if you want it to be a little bit cheerier, if you want it to be a little bit whatever, and you can start imagining also it might be able to tune these things to your needs.

And they could be, you know, this moment I'm talking to the other friendly robot, so I'm going to be very just the facts, but now I'm talking to a human, I want to be however the company wants you to be.

So it's a very interesting space.

I think it's going to evolve very quickly.

Yes.

So we're actually building an entire framework for doing what we're calling generative computing.

It's like, let's treat it, let's treat the thing not as a friendly robot, but let's treat it as a computational device.

It's like, you know, I send code to the GPU.

The GPU does something that the CPU couldn't do.

You get back the result.

We have heterogeneous processes all around.

Like, what if the LLM itself was like a processor, but it was really good at manipulating meaning and ideas and semantics and language?

Then we could use that to do lots of different things, some of them which aren't in the model of, like, friendly robot, whatever.

So basically, with some of these adapters, you just call a function, the library takes care of all the details for you, and you get back an answer, which is a structured thing, as if you had been a Python function that you called, or any other programming language.

Showing 141–160 of 634 · page 8 of 32