OpenAI had stunned the world by releasing ChatGPT in November 2022, but it turns out that the researchers working on the technology were even more stunned at what they’d developed.
Jeff Clune, a former researcher at OpenAI, says that building AI felt like watching an AI intelligence arrive. His words paint a vivid picture of a small group of scientists who felt they held a world-altering secret, a sentiment that oscillated between exhilarating and terrifying. Clune had worked at OpenAI from January 2020 to May 2022, and now works at Google DeepMind.
Clune’s analogy captures the surreal feeling of knowing the world was on the precipice of a monumental shift, while life for everyone else continued as normal. “It’s kind of like you’re an astronomer,” he explained. “You’re looking at your equipment, your sensors and your computer readouts, and you and a handful of other people have the expertise to look at this complicated data and say, ‘Oh my gosh, aliens are on the way. They’re going to arrive on Earth in a couple of years.'”
The sense of uncertainty was palpable. “We haven’t fully figured out how fast they’re traveling and what their technology is, but they’re going to be here real soon, and that’s going to change everything,” Clune said. The fundamental unknown, he suggested, was the nature of this impending change. “We don’t know whether or not it’s going to change it for the better or worse. Whether it will be good aliens or bad aliens, will it completely cause global chaos or will they usher in a new utopian era of great technology and friendships and knowledge expansion? Who knows.”
This feeling of holding a monumental, yet disbelieved, truth created a sense of isolation. “Like us seven people, or us 70 people, we know it. But nobody else listens to us and believes that this is happening,” he recalled. “I was speaking to journalists then, and nobody was listening. Everybody felt we were crazy, but we in that room, we knew that the data were pretty… we were sure that the aliens are coming.”
Adding another layer of complexity to the analogy, Clune acknowledged the active role the researchers played in this arrival. “The analogy’s even weirder because we weren’t just watching the aliens coming, we were also giving them information. We were helping them come to us by making the technology, which is also kind of a weird dynamic.”
This cognitive dissonance extended into their daily lives. “You know that, and then you walk home for the day and people are pushing their babies in a baby stroller and they’re going to the supermarket. You look around, it doesn’t look like the world’s about to radically change. It totally feels normal, but rationally, you know the world is going to be wildly different soon,” he reflected. “And I still feel that way.”
Clune’s sentiment resonates deeply within the current AI landscape. While the release of generative AI tools like ChatGPT, Midjourney, and Sora has certainly woken the world up to the power of this technology, a chasm of understanding still exists. The conversations have now reached the highest levels, with prime ministers and global leaders discussing AI’s implications. Yet, as Clune notes, skepticism remains. “You still have skeptics saying, ‘Oh, you guys are making this all up, and you’re hyping it for your own valuations and it’s hitting a wall and nothing’s about to happen.'” This skepticism, however, does little to shake the conviction of those who have seen the data. “But I’m very confident that the world is going to be radically different very soon,” Clune asserts. This perspective from an insider serves as a stark reminder that as astounding as recent AI advancements have been, they may only be the faint, initial signals of a much larger, more transformative arrival. The “aliens,” it seems, are no longer a distant prospect but are beginning to unpack their bags, and the world is only now starting to grapple with what that truly means.