Joe Biden meets CEOs of AI companies: Key takeaways
What's the story
The rise of generative AI has fueled new hopes but also new fears. ChatGPT and its ilk have become the latest cause of worry for governments and lawmakers across the world. Amid the growing appetite to regulate the technology worldwide, CEOs of top AI companies, including Google and Microsoft, met with US President Joe Biden. Let's look at the key takeaways from the meeting.
Context
Why does this story matter?
Regulating generative AI has become an important agenda for several countries. In Europe, lawmakers have reached a preliminary agreement regarding the AI Act. UK's competition watchdog has started the review of AI models. These activities have brought the spotlight to the US. Therefore, the meeting at the White House has attracted a lot of attention from AI supporters and naysayers.
Risks
AI systems pose risks that we are unaware of
We have seen what chatbots like ChatGPT or Bing Chat are capable of. Despite their abilities, they tend to make mistakes. They are also prone to hallucination, which leads to false answers. These are some of the known risks of AI. With the technology set to become more advanced, there ought to be more risks awaiting us.
Current and future
Government asked CEOs to mitigate current and future risks
AI companies are yet to develop a robust mechanism to deal with the risks posed by their models. Biden and Vice President Kamala Harris stressed the need to mitigate current and potential risks to realize the benefits of AI. Harris told the CEOs they have a legal responsibility to ensure their AI products are safe to use and won't cause any harm.
Transparency
AI companies should be transparent about their products
One of the major criticism leveled against AI companies is they are not transparent about their products. During the meeting, the US officials talked about the need for companies to be transparent with policymakers. Transparency about AI systems is limited to algorithms so far. Many have been calling for transparency in the interactive and social side of things.
Measurability
Officials talked about validating AI's safety, security, and efficacy
AI models have been around for a while, but it is only now they have become popular. Therefore, we have no way to "evaluate, verify, and validate" the safety, security, and efficacy of AI systems. This was one of the key talking points of the discussion, which the White House called "frank and constructive." A process to determine these would help build trust.
Information
The administration wants CEOs to promote responsible AI
We have been hearing about "responsible AI" ever since AI systems became popular. During the discussion, the administration officials emphasized the role CEOs must play in cultivating responsible behavior. The government wants companies to ensure they are innovating responsibly and with adequate safeguards.
Crackdown
Administration asked companies to adhere to AI Bill of Rights
The Biden administration has tasked Harris to contain the evils of AI. The administration will spend $140 million to launch seven National AI Research Institutes focused on developing ethical and responsible AI. This will bring the total to 25. The officials also asked companies to follow the blueprint for the AI Bill of Rights and the AI Risk Management Framework.
Public evaluation
AI systems will be subjected to a public evaluation
The White House also obtained an independent commitment from leading AI companies to participate in a public evaluation at the upcoming Defcon 31. Hackers at the event will test LLMs built by Anthropic, Google, Hugging Face, NVIDIA, OpenAI, and Stability. They will assess how these AI systems align with the AI Bill of Rights.
CEOs
Companies are on the same page: Sam Altman
The meeting with Biden involved CEOs from rival companies who are fighting for the top spot on the AI food chart. According to reports, the CEOs did not display any opposition. In fact, they seem to be on the same page on regulations. "We're surprisingly on the same page on what needs to happen," OpenAI's CEO Sam Altman said.
Attendees
CEOs of Google, Microsoft, OpenAI, and Anthropic were present
The meeting at the White House was attended by CEOs of four top AI companies, including Google's Sundar Pichai, Microsoft's Satya Nadella, OpenAI's Altman, and Anthropic's Dario Amodei. Representing the US government were Vice President Harris, Biden's Chief of Staff Jeff Zients, National Security Advisor Jake Sullivan, and Secretary of Commerce Gina Raimondo, among others.