CNAS analysts Vivek Chilukuri, Bill Drexel, and Tim Fist discuss Biden's 50-page executive order on AI. They cover immigration and federal hiring, safety concerns with AI + bio, cloud access, training GPT5, open source, and the desire for an AI czar.
The executive order allows federal government officials to directly hire individuals with AI expertise, addressing the challenge of attracting and retaining AI talent within the government.
The order seeks to streamline immigration processes for foreign AI talent to work in the United States, aiming to attract and retain AI professionals by finding ways to ease visa challenges and provide more flexibility in hiring.
The executive order emphasizes AI safety and security, establishing criteria for identifying AI models that pose potential risks to national security and highlighting the need for responsible deployment and use of AI to protect public safety.
Deep dives
Expanded hiring opportunities for AI experts in the government
The executive order introduces a provision that allows federal government officials to directly hire individuals with AI expertise, bypassing traditional hiring barriers. This aims to address the challenge of attracting and retaining AI talent within the government, which is crucial for effective development and implementation of AI policies and programs. By lowering the barriers to entry, the government hopes to bring in experts who can contribute to the advancement of AI initiatives.
Streamlining immigration processes for AI talent
The executive order includes measures to streamline the immigration processes for foreign AI talent to work in the United States. It seeks to attract and retain AI professionals by finding ways to ease visa challenges and provide more flexibility in hiring. The order calls for a review of current visa policies and considers options to simplify and modernize the pathways for AI professionals to enter and work in the country. By addressing visa-related obstacles, the government aims to ensure that the United States remains a hub for AI talent.
Focusing on AI safety and security
The executive order places significant emphasis on AI safety and security. It establishes criteria for identifying AI models that pose potential risks to national security, especially in the domains of cyber and biosecurity. The order requires companies to report on foundational models and computational infrastructure. It also highlights the importance of developing regulations and guidelines to ensure the responsible deployment and use of AI in order to protect public safety and safeguard against misuse or unintended consequences.
Expanding international collaborations on AI governance
The executive order aims to strengthen collaborations with international partners on AI governance. It promotes engaging with allies and global initiatives to address AI-related risks and establish standards for AI development, deployment, and safety. The order highlights the need for coordinated efforts to steer AI governance in a direction that aligns with democratic values and enhances international security. By encouraging multilateral engagement, the United States seeks to shape global AI policies and promote responsible and ethical practices.
Setting the stage for future AI regulations and policies
The executive order lays the foundation for future AI regulations and policies. It establishes the importance of data collection, evaluation, and technical standards in shaping effective AI governance. The order prioritizes the need for comprehensive reports, guidelines, and assessments from various agencies to gather empirical data and address specific concerns in areas such as AI research, ethics, and usage. By focusing on these key aspects, the government aims to inform future decision-making and ensure the responsible and successful development of AI technologies.
Biden just dropped a 50 page executive order that's going to make the world safe for AI, hopefully? To discuss the sprawling EO we've brought on three CNAS analysts, Vivek Chilukuri, Bill Drexel and Tim Fist.
We touch on:
Immigration and federal hiring
If AI + bio can ever be a safe thing
What's going to happen to cloud access
What are the hoops you'll need to jump through to train GPT5 and whether they're enough
What to do about open source
Why Jordan just wants to be an AI czar
RSVP to the Los Angeles meetup! https://partiful.com/e/SgjdajUSrD1aEOOrVgXk
Outro music (yea I was not going to impose Devo on you all): The B Tune by Bela Fleck https://open.spotify.com/track/6h6vvG1t4xtfP9lkOKzBTv?si=6ec3f32629bd46a2