Ahmed El-Kishky

So making sure that the ML is still stable, that the software is still great.

Um,

Making sure, you know, we have a really good way to interact with what we're trying to solve.

Because the models themselves, you know, we would like them to, you know, if they're working on something with chemistry, like it should, you know, get some feedback from like running a chemistry experiment.

Yeah.

So these kinds of things are, I think, the next sort of frontier.

It just, you know, makes a call to it with some text.

And yeah, the outputs from the tool just gets put back into the context.

And so now it has in its context, oh, this is what happened when I used this tool.

And so it can just like

uh use that to continue making decisions um like if you need to make a decision in a situation you need to you know have in your context what's the uh yeah what what the state of the world is the world can be your personal computer if it's like you know um it's trying to look at your directories and see what's in the specific folder um that would be sort of put back into the context and now it knows what's there and it knows hey you need to make like an edit to this file for example

Like looking at chess, you know, AI has surpassed many chess players.

I mean, all of them actually at this point.

Yeah.

But chess is still incredibly popular.

There is...

a tremendous amount of value in being a competitive programmer, someone that tackles these algorithmic problems.

And just because AI can do really well, that doesn't take from it.

There's a lot of valuable learnings that people get from going through that journey.

You get the ability to think really deeply to solve a hard problem.