Ben Gilbert

This is my favorite David Rosenthal-ism.

Everyone talks about 2022 onward as the AI era.

And I love this point from you that actually for anyone that could make good use of a recommender system and a classifier system, basically in a company with a social feed, the AI era started in 2012.

Yes.

This is Fei-Fei Li's thing from Stanford.

Yes, I think that's right.

Can't use it for much in a production setting when a quarter of the time you're wrong.

It's 40% better than the next best.

Yes.

On a relative basis.

Yes.

And why is it so much better, David?

What did they figure out that would create a $4 trillion company in the future?

That retail for $1,000.

And this is how AI research tends to work, is there's some breakthrough that gets you this big step change function.

And then there's actually a multi-year process of optimizing from there, where you get these kind of diminishing returns curves on breakthroughs, where the first half of the advancement happens all at once.

And then the second half takes many years after that to figure out.

But it's rare and amazing and...

It must be so cool when you have an idea, you do it, and then you realize, oh my God, I just found the next giant leap in the field.

Yes.