Washington DC.- The director of the Artificial Intelligence (AI) company that develops ChatGPT told the United States Congress that government intervention will be essential to mitigate the risks of increasingly powerful systems.
“As this technology advances, we understand that people are excited about how it could change the way we live. So are we,” said Sam Altman, CEO of OpenAI, at a Senate hearing.
The businessman proposed the creation of a US or global agency that would grant licenses to the most powerful AI systems and have the authority to revoke them and ensure compliance with security standards.
What started as panic among educators over the use of ChatGPT to cheat on schoolwork has broadened into more general concerns about the ability of the latest generative AI tools to deceive people, spread falsehoods, violate rights protections copyright and disrupt some jobs.
And while there are no immediate signs that Congress will draft sweeping new AI regulations, as European lawmakers are doing, societal concerns brought Altman and other tech CEOs to the White House earlier this month and have led to US agencies to pledge to crack down on harmful AI products that violate existing civil rights and consumer protection laws.
Sen. Richard Blumenthal, a Connecticut Democrat who chairs the Senate Judiciary Committee’s privacy, technology and law subcommittee, opened the hearing with a recorded speech that sounded like him but was actually a voice clone trained on the speeches on the full of Blumenthal and reciting the initial comments written by ChatGPT.
The result was impressive, Blumenthal said.
“What if I had asked him and provided an endorsement for the surrender of Ukraine or for the leadership of (Russian President) Vladimir Putin?”
In general, the tone of the senators’ questioning was polite Tuesday, in contrast to previous congressional hearings in which technology and social media executives faced harsh questions about the industry’s failures to handle the privacy of data or countering harmful misinformation. In part, this was because both Democrats and Republicans expressed an interest in seeking Altman’s experience to avoid problems that have not yet occurred.
Blumenthal said AI companies should be required to test their systems and disclose known risks before launching them, expressing particular concern about how future models could destabilize the job market.
Altman mostly avoided elaborating on his worst fear about AI, except to say that the industry could do significant damage to the world, and if it fails, it could do so in a major way.
He later proposed that a new regulatory agency impose safeguards blocking AI models that can “self-replicate and self-exfiltrate in nature,” hinting at futuristic concerns about advanced systems that could manipulate humans into relinquishing control.
Also testifying were Christina Montgomery, IBM’s director of privacy and trust, and Gary Marcus, a professor emeritus at New York University, who was part of a think tank that urged OpenAI and other technology companies to pause development of newer models. powerful for six months to give society more time to consider the risks. The letter was a response to the March release of the latest OpenAI model, GPT-4, described as more powerful than ChatGPT.
#regulated #ChatGPT #boss