From The Information
California’s Proposed Regulation Would Strengthen, Not Harm, AI
By AnnaLee Saxenian
Silicon Valley’s success has spawned some pernicious myths, most notably the glorification of the lone entrepreneur working in a garage, unfettered by government oversight. A closely related myth holds that regulation always jeopardizes innovation and technological progress, because government can’t wisely oversee technological change.
It’s an attractive story, but it’s not true. From the earliest days of the semiconductor industry to the rise of the internet, public sector support for technology has been crucial.
California’s SB 1047 establishes a light-touch regulatory regime, including a modest level of safety testing for the largest and most advanced generative artificial intelligence models—those that are trained on over 10^26 floating-point operations per second of computing power at a cost of more than $100 million and that could plausibly cause more than $500 million of damage. The bill would require developers of these models to take reasonable steps to reduce the risk of potentially catastrophic harms...
AnnaLee (Anno) Saxenian was previously Dean of the School of Information and she holds a joint faculty appointment in the School of Information and the Department of City and Regional Planning at UC Berkeley.