Ep15 - AI in Pharma: From Tool to Teammate with Darshan Kulkarni

Now, I know in the past and we've had a lot of conversations about legal regulatory compliance things.

You said that most legal vices either legally sound, but impractical or practical, but but risky.

What's an example of where you've had to reconcile that tension?

I think AI is a great place to talk about.

Right now, everyone's talking about AI.

A really good example of it right now is where everyone wants to use AI to make their job easier.

And that's totally fair, totally appropriate, except when it's done wrong.

For example, right now, if I tell someone you can't use AI, that's the most conservative legal advice I can give.

But let's be honest, no one's going to follow me on that.

On the other hand, if I say use AI all you want, what you get is a case that literally popped up, I want to say, three weeks ago.

This was a Judge Rakoff decision, I want to say in the third circuit.

Essentially what it was, was the lawyer told his client, That what I want you to do is I want you to organize your thoughts, give them to me.

So what the client does is he goes out there and he writes out all of his thoughts and then runs it through Claude.

And Claude was one of the free Claudes that he had, which was great.

He was like, oh, perfect.

Here's my organized thoughts.

The other side found out.

And they said the moment you use Claude, it was a disclosure to other people.

Because theoretically, Claude is not private.

So the moment it was a disclosure, you lose attorney-client privilege.

Now, that's bad.

Losing attorney-client privilege is terrible.

But here's why I think it matters in the life sciences.

In the end, the only thing life sciences cares about is your intellectual property.

The moment you have a situation where your intellectual property is discoverable or not otherwise protectable, you have a major problem on your hands.