California lawmakers have passed SB 1047, a significant AI safety bill requiring safety testing and third-party audits for advanced AI systems costing over $100 million. While it aims to balance innovation and safety, it faces opposition from some industry leaders and academics who argue it may hinder smaller firms and academic research. The bill emphasizes accountability for AI developers but raises concerns about its potential chilling effect on innovation. Advocates claim it is a necessary step towards ensuring safety in AI technology. Governor Gavin Newsom has not yet stated his position, and he is facing pressure from the tech industry and advocates for regulation.
The Office of Management and Budget (OMB) will soon release guidance to assist federal agencies in acquiring artificial intelligence technologies following feedback from various stakeholders. This memo aims to align agency contracts with best practices outlined in previous OMB guidance and is expected by September 24. Additionally, OMB has initiated AI training for employees involved in policy and acquisitions to enhance understanding and effective use of AI in government. A bipartisan bill is also being drafted to require risk assessments for AI technologies before purchase and use.
In Cheyenne, Wyoming, Victor Miller’s campaign to elect an AI chatbot named VIC as mayor ended in defeat. Miller received only 327 votes against incumbent Mayor Patrick Collins, who secured 6,286 votes. Miller described his campaign as historic and aimed to promote AI governance through a new organization, the Rational Governance Alliance, advocating for AI decision-making in public office. Despite the loss, Miller remains committed to advancing the role of AI in governance, arguing it could lead to more efficient and unbiased decision-making.
ESPN has begun publishing AI-generated sports recaps for women’s soccer and plans to expand to other sports, claiming this will allow human writers to focus on more in-depth reporting. However, initial recaps lack nuance and have missed significant moments, raising concerns about the quality of AI-generated content. One of the National Women’s Soccer League stories failed to mention the significance of one player’s final game and the emotional moments that happened, something ESPN waved at with a later update to the story.
Why do we care?
I’m starting with the legislation because it’s been sitting on the Governor’s desk for a while. It was passed on August 29… but you’ll note it’s September 12, and we have no updates. He has until later this month to decide. I had held off until it passed, but I think the delay is part of the story and why we care.
Beyond that, there’s some fundamental mismatch in use cases here. The public has spoken in Wyoming—no thank you to having a chatbot run the city. This should be obvious—use cases fail without a human in the loop. See the ESPN story. There is a subtlety to stories that can be critically missed. The key is quality – AI is an option if your goal is to provide something versus nothing. If you want to deliver quality, AI alone probably isn’t.