Skip to main content

The messy roll-out of artificial intelligence chat programs has not necessarily been a bad thing, said a leader of one of the companies leading the way with the new technology.

At the recent TED2023 conference in Vancouver, Greg Brockman, co-founder of OpenAI, said allowing public concerns to arise at early stages of ChatGPT-4′s development was better than creating it fully in secrecy.

“Confront reality as it lies,” Mr. Brockman said of this open-book approach. “Let reality hit you in the face.”

Open this photo in gallery:

Greg Brockman speaks onstage during the Dropbox Work In Progress Conference at Pier 48 on Sept. 25, 2019 in San Francisco.Matt Winkelmeyer/Getty Images

ChatGPT-4 is a large language model with the ability to write code, create images and give human-like responses to text inputs.

Mr. Brockman told the crowd at the annual event – which was this year centered around discussions of the impacts of new technologies – that companies have the option of vetting the safety of their products internally and then releasing everything at once. But this, he said, could create a technological “overhang,” where too much AI technology is released into the world all at once, without the checks and balances that outside experts, developers and the public provide when its instead available in stages.

Mr. Brockman add that releasing AI technology “incrementally” allows companies to use feedback to make corrections as they arise.

Another good reason to let the public try out new products, he said, is to allow people to become literate in emerging AI technology in order to “decide what we want from it.”

Mr. Brockman was joined on stage by Chris Anderson, the head of the conference and curator of TED. Mr. Anderson provided a counterpoint, arguing that releasing AI technology like ChatGPT-4 so early could force competitors to follow suit.

“It’s already out there,” said Mr. Anderson, adding that if something bad happens there may be no going back.

Concern over the rapid advancement of AI technology was underscored last month in an open letter authored by the Future of Life Institute, an organization of experts in science, research, business, medicine and engineering, aiming to “steer transformative technology towards benefiting life and away from extreme, large-scale risks.”

The letter, which was signed by over 26,000 people – including Tesla CEO Elon Musk, Apple co-founder Steve Wozniak and Skype co-founder Jaan Tallinn – calls for a temporary pause of “at least six months” to any artificial intelligence more powerful than OpenAI’s ChatGPT-4.

The institute hopes the pause will allow AI experts to “develop and implement a set of shared safety protocols for advanced AI design and development.”

At the conference, Mr. Brockman demonstrated the ChatGPT-4 application onstage by asking it to suggest a post-presentation meal, and then create an AI-generated picture of the meal, craft a tweet to show it off, and order the necessary ingredients on Instacart, the online grocery delivery service.

“This is all a live demo … I actually don’t even know what we’re going to see,” said Mr. Brockman. Then, just seconds, ChatGPT-4 generated everything he asked of it in real-time, for the audience to see on a large screen behind him.

As the AI-generated tweet went live on Mr. Brockman’s Twitter account, the audience applauded.

The advancement of AI is “every bit as significant as the creation of the internet,” Mr. Anderson said to media on the first day of the conference. “We better get it right.”

Follow related authors and topics

Authors and topics you follow will be added to your personal news feed in Following.

Interact with The Globe