The ChatGPT logo shown on a screen in an office in Washington, DC on March 15, 2023 (Getty Images)
OpenAI, the AI firm behind ChatGPT, has warned that the advent of artificial intelligence that surpasses humans is inevitable.
Artificial general intelligence, also known as superintelligence, has been theorized by philosophers and academics for decades, though rapid progress in recent years means we may now be on top of it, according to senior OpenAI figures has warned.
In a blog post published Wednesday, OpenAI’s Sam Altman, Greg Brockman and Ilya Sutskever said AI superintendence will be unprecedented in its power — both positive and negative.
“Given the picture as we see it now, it is conceivable that within the next ten years, AI systems will exceed expert skill levels in most domains, and be as productive as one of the largest corporations today.” Will do the activity,” the post said.
“In terms of both potential upside and downside, superintelligence will be more powerful than other technologies humanity has faced in the past.”
OpenAI has set out three ways humanity can navigate the advent of superintelligence, though cautioned lawmakers and regulators against trying to prevent it.
“We believe it would be unconscionably risky and difficult to stop the construction of the superstructure,” the post warned.
“Because the benefits are so tremendous, the cost of making it goes down every year, the number of actors making it is growing exponentially, and it’s naturally part of the technological path we’re on, to prevent would require something like a global monitoring system, and even that is not guaranteed to work. So we have to fix it.”
Mr Altman appeared to face questions from US senators during a congressional hearing last week about the risks posed by advanced artificial intelligence.
OpenAI CEO Sam Altman appears before the Senate Judiciary Subcommittee on Privacy, Technology and the Law on May 16, 2023 (YouTube/ Screengrab)
The 38-year-old told the Senate Judiciary Subcommittee on Privacy, Technology and the Law that he believed AI-powered chatbots like ChatGPT were a “significant area of concern” and that rules and guidelines were needed to prevent abuse.
“There’s no way to put this genie in a bottle. Globally, it’s exploding,” acknowledged Democratic Senator Cory Booker.
Introducing licensing and testing requirements for AI development would be a possible way to prevent AI harms such as election manipulation, Mr. Altman said.
According to OpenAI, one possibility is through the creation of a US licensing agency for AI called the Office for AI Safety and Infrastructure Security (OSIS).