Jonathan Schaeffer

What happens when you put an AI pioneer with more than 40 years of experience in the same conversation with two podcast hosts who have some pretty strong opinions about artificial intelligence? Well, as it turns out, you get a pretty lively conversation.

Our guest today is Jonathan Schaeffer, a computer scientist who helped teach computers to master checkers and poker, and who now believes that much of the AI industry has gotten something fundamentally wrong.

His answer is kind, an AI product that can run entirely on your own computer without sending your confidential information to the cloud.

We also got into AI hallucinations, the possibility of rogue AI agents escaping their safeguards, and Jonathan's rather alarming prediction about what it may take before governments take AI safety seriously.

I'm Mildred Gardner.

And today we're joined by someone who's been working in artificial intelligence since long before most people had ever even heard the term.

Jonathan Schaeffer is a computer scientist, professor at the University of Alberta, and one of the pioneers of modern artificial intelligence.

It's exciting to have him here.

And I am your other co-host, Spencer.

And today we are here with Jonathan Schaeffer, um, a true pioneer in AI.

He's been around for a long time, and he led a team that built Chinook, an AI system that built, uh, that won a world championship and helped solve checkers completely, um, which is a, a big deal.

Um, so he's had over four decades in AI, uh, so we brought him on today to talk to us about current developments.

What's nice about what we do is every fact in the answer is cited back to your data with the source.

So it might be that we s- include something in the answer, and we'll point to you and say, "It's on page 25 in this document.

Click here, and it'll bring up that document, page 25." Or, "Jonathan Schaeffer said this," and you click here, it takes you to 18 minutes and 25 seconds into a video, and then you hear Jonathan Schaeffer saying that.

Yeah.