AI@GSB Perspectives: AI & Politics with Anthropic's Andy Hall & writer Jasmine Sun
Stanford Graduate School of Business
Election context 0:04
The speakers open by noting the big election week and then steer the talk toward AI. Andy Hall is introduced as a Stanford professor who studies politics and AI, and Jasmine Sun as a writer who has reported on data center buildout and the people around it. The first question asks whether public feeling about data centers is the same as feeling about AI itself, or whether those are different things.
Local anger and national doubt 2:00
Sun says the local fights are usually about water, electricity, noise, truck traffic, and the feeling that companies made decisions without real public input. People in Wisconsin and Michigan rarely raised AI risk, surveillance, or jobs unless she brought those up. Hall agrees that elections are made of many separate choices, not one clear story. He says this election showed swing voting and some support for more centrist candidates, but he is unsure that local anger at data centers will turn into broader support for AI safety.
Political effects remain hard 15:01
The speakers say it is still very hard to explain how the economy, labor market fears, and AI shaped the vote. The economy is the strongest predictor of election results, but even that is messy because GDP can look good while cost of living does not. They think AI will keep growing in policy importance because politicians are watching it and the technology is moving fast.
Data centers and perception 17:00
On data centers, they are skeptical that better branding or more ads will solve the backlash. Once a project becomes a polarized symbol, the fight is really about power, trust, and who decides for a community. They suggest state-level deals may work better than local ones. They also note a gap between heavy users who love AI and most people who see it as a useful but not essential tool.
Biased answers and local data 30:01
Chatbots can seem left of center because they learn from the open web, where young politically active people are often left leaning. The harder question is not only how to make answers less biased, but what you want political chatbots to do at all. Japan shows the problem. When mainstream media blocked AI access but a Communist Party paper stayed open, chatbots looking up voting advice often returned that fringe source and told users to support the Communist Party.
Scraping and political reach 34:30
That same fight shows up in other places. Political parties will try to get their material ranked high for AI answers, much as they once did for Google search. Similar choices came up at Substack. The default there still allows AI scraping, though writers can turn it off. The choice was made so writers could shape the models people use, even if many writers wanted a blanket ban and disliked the idea of their work being scraped.
Election politics 44:01
Public distrust of AI is shaping politics. Moderate candidates have trouble defending continued AI development, and a pause may become the middle option in the 2028 Democratic race. The talk also links data center fights to AI worries, especially about jobs, though the speakers think most voters are driven more by quality of life, corporations, and democratic process than by AI names like ChatGPT or Claude.
Pauses and deal making 48:00
The discussion turns to moratoria on data centers. One view is that they are meant to slow projects and force better community terms, not just block them. The speakers say recent deals have improved for towns, after earlier ones left communities in a weak position. State leaders can standardize terms, set fair tax and benefit rules, and reduce the gap between company lawyers and local officials.
AI-generated summary. It can be wrong or incomplete - check anything that matters against the original.

