Jason Calacanis

If you look at the companies that use that software,

Those companies generate enormous revenues and enormous margins, and these products are in critical production workflows that underlie those revenues and profits.

That is just not true with AI today.

We have all kinds of claims, but we are still experimenting.

Why are we experimenting?

Because we know it's important, but we don't yet really know what to do.

You can't just slot this in to a critical workflow in healthcare and all of a sudden show up where if you make a misdiagnosis or if you make a mischaracterization of a procedure, you can get fined and go to jail.

The companies that are in healthcare don't do that.

If you're in financial services and you make a mistake about somebody's portfolio or you make a misallocation and you point to a model, you will get sued and you will be in trouble.

None of these things have transitioned from it's interesting, it's experimental to it's the core critical operational workflow.

It's interesting.

There will be a transition in revenue quality when that happens.

A great example of this is Amazon.

Why does Amazon issue an edict that says you cannot use this stuff inside of AWS unless a human now reviews and approves it?

Because what happened?

They had three or four SEV1 faults from a bunch of code that was written by agents that brought down AWS.

Now, look, I've told you, I love AWS for one reason, because it's hyper-reliable.

I hate AWS for the same reason, that hyper-reliability comes at enormous cost.

I pay it, but I pay it to never have a SEV1.

The reason they have 12 nines of accuracy is because it's humans and deterministic code that never fails.