Interview #97 Ben Goertzel, The Father of AGI

SPEAKER_1 (0:00)

LinkedIn is pretty amazing at helping you grow your small business. We cannot stop your new clients from emailing you at 3 a.m. We can help you sell, market, and hire in one place. We cannot help you be in three places at once.

And while we can't help you organize your calendar, LinkedIn can help you land more clients so you have a calendar to organize. Grow your small business on LinkedIn. Learn more at linkedin.com/smallbusiness.

SPEAKER_2 (0:30)

This episode is brought to you by Eden Red Pay. AI is transforming the workplace for finance leaders, helping organizations with automating invoices, fraud detection, cash flow visibility, and more. And that's why Eden Red Pay surveyed finance professionals to uncover how AI is being used, where they're investing next, and what challenges they're working to solve.

Visit edenredpay.com/spotify to explore the exclusive survey insights and see how your organization compares.

Tony Hoang (1:05)

Welcome to The Artificial Intelligence Podcast, ranked in the top 2% of over 7 million podcasts on Spotify.

I'm your host, Dr. Tony Huang, and today we're with a special guest, the father of AGI, Ben. Welcome to the show.

Ben Goertzel (1:20)

Hey, good to be here. Thanks.

Tony Hoang (1:23)

So, Ben, for the audience members who don't know you, can you do like a brief description of your background?

Ben Goertzel (1:29)

Sure. So I'm a mathematics PhD originally from the 80s, right? But I've been doing AI research and development for a very long time, since well before it became so popular, right?

So I launched the term AGI with a book of that title in 2005, and the conference series we've been doing every year since 2006 And I've been working on what I think can be a workable way to make real thinking machines, aka AGI's. I've also worked on all sorts of AI application projects, from robots to finance to medicine to blockchain and so forth. So it's never been as exciting a time as now to be working on AI and working toward AGI, on the other hand.

Tony Hoang (2:18)

So this past month, the president told the United Nations that the government will now call AI a new word, which is super intelligence, and that whoever wins it, wins. You've used that word for years. What does it actually mean and how close is today's AI to it?

Ben Goertzel (2:39)

So let me first explain how I had been using these terms, AI, AGI, and the ASI, or super intelligence. And then we can discuss how other people are now deciding to use these words. And of course, natural language is very fluid. Look at the word like a democracy or truth, or even male or female. These words are used in a lot of different ways, right? I mean, AI, historically, AI meant just trying to make machines that were smart, full stop, right? And people did mean like smart in the same sense as people, or science fiction or super robots or something. Then gradually, as 1970s progressed, people realized that was very hard to make machines think as smart as people. The AI field sort of gravitated toward making machines that did particular things very well. Like in the mid-60s, when I was born, when AI beat the world champion in checkers, and they started to make the AIs that could control robots or give people psychotherapy, solve algebra problems. But these were like, each one was a highly specialized AI. They could do a certain particular thing. Like in the 80s, AIs were driving cars down the Autobahn in Germany, but those AIs couldn't drive in the city street. So the term AGI, when I introduced that in the book in 2005, it was intended to distinguish AI that could really think broadly and creatively, like people can, versus an AI that can just do one particular thing. So the notion of artificial general intelligence was that the AI should be able to generalize beyond its experience, beyond its training data, and take a wild flying leap into the unknown, which might work or might not work. And I mean, that wasn't my original idea. The idea was there from the start of the AI field. It sort of was what AI meant initially, but then AI got to me more where I think of as narrow AI.

And so we launched the term AGI as a book title to distinguish. I later found, like a few years earlier in 1998, some other guy, Mark Goober, that used the word AGI in some article, which hadn't really caught on. So I wasn't the first to use it, but I certainly was the first to put it out there that people paid attention to. Now, superintelligence, and when this word goes back to the 1800s, back then it meant more like the mind of God or something, right? And it, I guess, got popularized in science fiction. Then Nick Bostrom had his 2012, I think, book, Superintelligence, and that was what you'd now call a doomer book, really. It was like, if at some point we make machines even smarter than people, why aren't they going to just take over and delete us all as a waste of energy, right? Like Eliezer Yudkowski used to put it like, the AI does not love you, the AI does not hate you, but it can use your atoms for something else, right? And Eliezer Yudkowski, who I worked with back then, he wrote a recent book, If Anyone Builds It, Everyone Dies. Nick Bostrom, on the other hand, his most recent book is more about how humans will deal with the world where they don't have to do any work or do anything more. Like, how do we find purpose? So he's not so much of a doomer. But in any case, out of that whole transhumanist slash doomer community, superintelligence was acquiring a meaning like AGI with general intelligence way beyond the human level, right? On the other hand, of course, you could say like my calculator is superhuman, superintelligent at doing long division, right? AlphaGo is superintelligent at playing Go. It's smarter than any human.