AGI pioneer Ben Goertzel thinks the AI race is too narrow

Ben Goertzel always believed the world would eventually take artificial general intelligence (AGI) seriously. What frustrates him is the commercial direction AI has taken. He sees a field full of possibilities but an industry concentrating much of its resources on just one part of it.

Goertzel helped popularize the concept of AGI more than two decades ago, coediting the book Artificial General Intelligence and helping establish a dedicated research conference. While AGI has become a central goal of the AI industry, his approach to achieving it remains outside the mainstream.

Today, he runs SingularityNET, which develops open, decentralized AI technology, as well as BGI Labs, which builds proprietary products on that infrastructure. He also leads the AGI Society, a nonprofit community dedicated to studying artificial general intelligence.

Goertzel says cryptocurrency funding has helped finance his research. Rather than relying primarily on scaling large language models (LLMs), his approach to achieving AGI combines those models with logical reasoning and long-term memory to develop AI systems that can continually learn from experience.

In this interview, Goertzel explains why he believes the AI industry’s fixation on LLMs could hold back progress toward AGI, what today’s systems are missing, and why, as he approaches 60, being the first to achieve AGI matters less to him than seeing it built.

This interview has been edited for clarity and brevity.

You’ve worked on AGI for decades. What has it been like to watch it become a central goal of major AI labs?

We always knew this was going to happen. It was a matter of when and exactly how. Back then, AGI was an unpopular thing to work on. You couldn’t get a job doing it. Only a handful of university departments were open-minded enough that you could give a seminar on it.

It was more like time travel or psychic power – something you talk about in the bar over drinks rather than in the seminar room. But it was fairly obvious that eventually breakthroughs would happen, and then everyone would take it seriously.

Was ChatGPT the breakthrough that changed the global course for AGI?

ChatGPT was a big moment, but there was a series of them.

In 2006, I was working with Peter Thiel on his hedge fund, Clarium Capital. We met at his place in the Four Seasons in San Francisco and talked about AGI for the afternoon.

He understood it was coming, he knew it was real, but he didn’t opt to invest in my project then. He wasn’t sure if it was coming five, 10, or 15 years down the line.