Sitting in a dark room with almost two
thousand technologists, scientists, entrepreneurs, educators, and artists, the ordinarily passionate audience was a little anxious. They knew there was the elephant in the room. TED leans in favor of
technology driving change in a positive direction. But when it comes to artificial intelligence, the feelings were deeply mixed.
Chris Anderson, the head of TED, brought the group’s enthusiasm and fears front and center in a session of speakers carefully curated to discuss fast-moving innovations and looming risks.
The advent of AI is “as significant as the advent
of the internet” said Anderson, and “the most important conversation we will be having. For better or worse, we’d better get it right.”
To open the session,
Anderson brought OpenAI cofounder Greg Brockman onstage. Brockman showed off Chat GPT and some of its as-yet-unreleased wizardly tools. “It’s amazing, but it’s really scary,”
said Anderson after Brockman’s TED talk ended, and the two began a Q&A session.
Anderson: “So here is one of the big fears then that arises. It’s fundamental to
what’s happening here, that as you scale up, things emerge that you can maybe predict with some level of confidence -- but it's also capable of surprising you. Why isn't there just a huge risk
of something truly terrible emerging?”
But Brockman sees OpenAI’s decision of testing the public as the only way to achieve big change. “We really were afraid that the
number-one thing people were going to do with it was generate misinformation, try to tip elections. Instead, the number-one thing was generating Viagra spam.” The audience laughed nervously
here. You can watch the whole interview here.
advertisement
advertisement
Moving on, TED brought to the stage Gary Marcus, a cognitive scientist, author,
and entrepreneur. Marcus pulled no punches, saying we need an international AI regulatory body. “A global, neutral nonprofit with support from governments, big business, and society is an
important start.” He wants more than governance, he wants to integrate ChatGPT’s brute statistical power with a more trustworthy, logic-based system. Coming after Brockman’s talk,
which was essentially “move fast and break things” 2.0, Marcus' words took on a haunting seriousness. This is more than theory.
Back on stage, Anderson noted that the
next speaker had been invited just days before to add an even more urgent voice to the discussion. The controversial decision theorist Eliezer Yudkowsky took the fear factor up a notch, noting that
superintelligent AI could probably kill us all.
Is an obedient, even benevolent, AI of superhuman intelligence possible? Yes, Yudkowsky said, but inscrutable large language models like ChatGPT
are leading us down the wrong path. By the time the world realizes this, he thinks it may be too late. "I suspect we could figure this out with unlimited time and unlimited retries," he said,
but "we do not get to learn from our mistakes and try again. I think we all end up dead.”
TED’s strength is the diversity of its community, and its ability to amplify issues
by sharing its talks on a global platform. "The conversation can't just be had by technologists," Anderson said. "And it can't just be heard by politicians. And it can't just be held by creatives.
Everyone's future is being affected. And so, we need to bring people together.”