1. OpenAI CEO Sam Altman has called for a licensing regime for powerful AI models, which would represent a dramatic expansion of federal power over the AI sector.
2. IBM and the European Union have proposed more conventional approaches to regulating AI, focusing on overseeing the use of AI in high-stakes domains like criminal justice, hiring, and medicine.
3. Congress should fund efforts to better understand the potential harms from AI and create an agency to investigate cybersecurity vulnerabilities in real-world systems.
The article argues against rushing into regulating AI and highlights the lack of clarity in OpenAI CEO Sam Altman's proposal for a new licensing regime for powerful AI models. The author suggests that while it may make sense to have a new regulatory framework for AI, there is a risk of getting it wrong given how fast the technology is changing. The article also discusses IBM's "precision regulation" approach and the European Union's proposed AI Act, which focuses on regulating consumer-facing applications of AI.
The author notes that Altman's proposal would represent a significant expansion of federal power over the AI sector and questions whether such a system could work given the lack of detail provided. The article also highlights potential issues with regulating providers of foundation models, which could create a schism in the AI world and limit access to cutting-edge US models.
The article raises concerns about job losses due to AI but notes that predicting the impact on jobs is difficult. It also discusses potential harms from AI-generated misinformation and deep fake images, audio, or video that could deceive voters and influence how they vote.
The author suggests that Congress should fund efforts to better understand potential harms from AI and create an agency to investigate cybersecurity vulnerabilities in real-world systems. However, the article questions whether Congress's sense of urgency on this issue is justified and suggests waiting to see how AI technology evolves before passing major bills to regulate it.
Overall, the article provides a balanced view on regulating AI but leans towards caution in rushing into regulation without fully understanding its potential impacts. The author acknowledges potential risks from AI but also highlights limitations in current proposals for regulation.