Private Equity and AI: From Hype to Portfolio Execution

So James and Nathan, briefly as we kinda get to the next core part of our conversation, what was AI like five to 10 years ago? Was it even a word?

It was definitely a word.

It was artificial intelligence.

Been around since, I mean, AI was at the Dartmouth conference in 1956, was the birth of artificial intelligence, so it's been around for a very long time.

Just didn't work very well [laughs] for a very long time.

I mean, 10 years ago we had a lot of conversations about, is this really AI? And there was plenty of examples of things that I, I consider to be really AI in the world at the time.

I mean, things like Google Maps is an example of AI.

It's a way to have a probabilistic outcome depending on a bunch of calculations to get you to do something that you didn't think was possible with computing at the time.

There was examples of it around.

You'd also have to find a, a data set, work out what the ground truth answers were to a data set, make sure you had it all, feature condition, the data had to be groomed really, really well.

You had to not over fit models.

These are all things we don't talk about much anymore.

I know they do on the, the really deep end of the AIs world.

That all changed just overnight when ChatGPT came out two years ago.

It's a Gutenberg moment.

So crazy to see an LLM nail it like ChatGPT did two years ago.

We did have LLMs previous to that, and to an extent, that's how Alexa works and, and other voice activated things, but it was nowhere near as easy and good as ChatGPT.

It's similar to how the iPhone changed things up in 2007 as well.

But, you know, it just unlocked a whole bunch of opportunity for everybody overnight.

So it was there, it just was a lot harder and didn't work really well.

Nathan, from your perspective, if you go back to your organization and you looked at all the startups five years ago, how many of them were doing deep AI kind of initiatives versus application software?

That's been the biggest shift, I would say.

And we collect data on everyone who comes to Venture Cafe and participates in one of our programs, the reason being is that we wanna know who's coming, what's their areas of interest, what's their stage of business, what technology enablers are they focusing on, are they utilizing.

And eight to 10 years ago, or five years ago even, when AI was referenced as a technology enabler that's being utilized, obviously there was smaller amounts that were indicating that.

But even those that were, it was one of just a mix of all of the other enablers.