Drive Fast, Learn Faster
Speed vs control, racing legend's startup lesson.
David SubarGuest
Alex BelotskyHostKuba FiatkevichHost
So let's say you're on the board of a company, pretty much like almost all companies now looking to deploy some form of AI, whether it's a core of their product or a standalone service in some way.

What do you recommend? What would you tell them? to the execs that are bringing this on.

And what would be some of the first things that you would make sure that they're covering?

Are you building AI into the product because that's cool? Are you building it because it's adding value? That's the first question I would ask.

Then I would ask, is it differentiated value? Is it sustainable? So let's say you're just building in AI you know, some arbitrary LLM that anybody can get.


Okay, is it, you know, is it adding value? Is it differentiated? Does it create some kind of a moat? There could be, by the way, an argument is our competitors are all doing this and it creates, and the customers want it.

But if you're putting it in, for a new feature is what value are you creating for the consumer? This is not necessarily a different question than we had before AI.

But I think that there's a little bit more rush to add some AI to a product because it's cool

because it might feel like it artificially inflates the market cap of the company.

Brand, well, brand can be sustainable, but not just, I'm putting this thing in it because it's cool, because anybody can do that.

Is there something that you would be cautious about or looking out when deploying or pushing out AI into production?

One is, and this, the first one goes to your field is, how are we going to know that this performs as we expect it to? That the thing about AI kind of any stochastic system is it's hard to test the boundaries because the boundaries might be in the middle.

Like it might perform very well on 15 test cases and then just perform really badly in the test case in the middle and then perform well.

So For instance, there was, you know, put together an arbitrary generation of video or arbitrary generation of an image.

So let's say you're on the board of a company, pretty much like almost all companies now looking to deploy some form of AI, whether it's a core of their product or a standalone service in some way.

What do you recommend? What would you tell them? to the execs that are bringing this on.

And what would be some of the first things that you would make sure that they're covering?

Are you building AI into the product because that's cool? Are you building it because it's adding value? That's the first question I would ask.

Then I would ask, is it differentiated value? Is it sustainable? So let's say you're just building in AI you know, some arbitrary LLM that anybody can get.


Okay, is it, you know, is it adding value? Is it differentiated? Does it create some kind of a moat? There could be, by the way, an argument is our competitors are all doing this and it creates, and the customers want it.

But if you're putting it in, for a new feature is what value are you creating for the consumer? This is not necessarily a different question than we had before AI.

But I think that there's a little bit more rush to add some AI to a product because it's cool

because it might feel like it artificially inflates the market cap of the company.

Brand, well, brand can be sustainable, but not just, I'm putting this thing in it because it's cool, because anybody can do that.

Is there something that you would be cautious about or looking out when deploying or pushing out AI into production?

One is, and this, the first one goes to your field is, how are we going to know that this performs as we expect it to? That the thing about AI kind of any stochastic system is it's hard to test the boundaries because the boundaries might be in the middle.

Like it might perform very well on 15 test cases and then just perform really badly in the test case in the middle and then perform well.

So For instance, there was, you know, put together an arbitrary generation of video or arbitrary generation of an image.
Every episode on Radar is fully transcribed, speaker-labeled, and rich with metadata. Here is a taste of this one. Try Radar for free to see the rest.
3 of 5
Drive Fast, Learn Faster
Speed vs control, racing legend's startup lesson.
AI Eating Software Revelation
The next predator in technology's food chain.
Boards Are Ignorant Confession
Why the board might still be clueless.
+2 more clips · 3 min 39 sec of audio in all
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
All 5 clips — the highlight moments, each cut as its own audio, with a title and a speaker
All 11 segments — the transcript broken into labeled sections, every ad read marked
All 23 topics — jump to every other episode discussing the same subject
Every related episode — other shows Radar links to this one
No account is needed to search Radar.