Prabhakar was instrumental in passing an executive order on AI in 2023. The executive order sets rules for tech companies to make AI more secure and transparent (though it relies on voluntary participation). Before joining President Biden’s Cabinet, he held a number of government positions, from campaigning for domestic semiconductor production to heading DARPA, the Pentagon’s famed research arm.
I had the opportunity to speak with Prabhakar earlier this month. We talked about the risks of AI, immigration policy, the CHIPS Act, public trust in science, and how everything will change under the Trump administration.
The change in government comes at a chaotic time for AI. Although Trump’s team has not offered a clear theory on how to handle artificial intelligence, many within the team want to see the executive order repealed. Trump made similar remarks in July, endorsing the Republican Party’s platform that the executive order would “stifle AI innovation and impose radical leftist ideas on the development of this technology.” Industry heavyweights like venture capitalist Marc Andreessen say they support the move. Complicating that story, however, will be Elon Musk, who has for years expressed concerns about AI doomsday scenarios and supported some regulations aimed at promoting AI safety. . No one knows exactly what will happen next, but Prabhakar has a lot of ideas about what has happened so far.
Read my conversation with Arati Prabhakar for her insights on the most important AI developments of the previous government and what will happen in the next one.
deeper learning
These AI Minecraft characters did all the strangely human things themselves
The video game Minecraft is becoming increasingly popular as a testing ground for AI models and agents. This is a trend that startup Altera recently embraced. Leverage large-scale language models (LLMs) to unleash up to 1,000 software agents at a time to interact with each other. Just by giving them nudges through text prompts and without any additional input from their human creators, they developed an amazing range of personality traits, preferences, and expert roles. Remarkably, they spontaneously made friends, invented jobs, and even spread religion.
Why this matters: AI agents can perform tasks, exhibit autonomy, and take initiative in digital environments. This is another example of how impressive and downright bizarre the behavior of such agents with minimal guidance from humans can be. The people working to introduce agents into the world have bold ambitions for them. Altera founder Robert Yang believes the Minecraft experiment is an early step toward a large-scale “AI civilization” with agents that can coexist and collaborate with us in digital spaces. “The true power of AI is unlocked when we have truly autonomous agents capable of collaboration at scale,” Yang said. Read more from Niall Firth.
bits and bytes
OpenAI is considering advertising