Scott Wiener is taking down the chatbot he launched to mock congressional opponent Connie Chan after receiving backlash from Pelosi and other Democratic lawmakers, acknowledging he \u201cclearly missed the mark.\u201d<\/p>\n
State Senator Scott Wiener announced Friday that he was deactivating the chatbot he created to mock San Francisco Supervisor Connie Chan, and he’ll also be removing the billboards promoting it, adding that the controversy had overtaken the issues he wanted to highlight in the congressional race, according to Bay Area Reporter<\/a>.<\/p>\n \u201cThroughout my career, I\u2019ve never been afraid to push the envelope, which means taking risks, sometimes succeeding, and sometimes missing the mark,\u201d Wiener stated.<\/p>\n Prior to Friday’s announcement, Wiener stood by the chatbot, maintaining that the site was clearly labeled as parody and that \u201cno reasonable person\u201d would mistake it for an official Chan campaign platform. <\/p>\n On Thursday, Speaker Emerita Nancy Pelosi criticized the bot, which was designed to mock SF Supervisor Connie Chan’s positions on issues including housing, public transit, education, and San Francisco\u2019s economic recovery.<\/p>\n Pelosi, who has endorsed Chan<\/a>, said the chatbot went beyond political satire after Chan\u2019s campaign prompted it about her accent and citizenship and received responses that invoked racial and sexist stereotypes, as Politico reports<\/a>. Pelosi said one response echoed \u201charmful racial tropes\u201d associated with Trump-era birtherism.<\/p>\n The letter Pelosi posted Thursday was also backed by Representative Zoe Lofgren, eight other Democratic women in Congress, San Francisco Supervisors Jackie Fielder and Chyanne Chen, and several local Democratic clubs.<\/p>\n \u201cWomen have spent generations fighting to have our own voices heard in the halls of power,\u201d Pelosi and the other signatories wrote. \u201cWe should not have to now fight against artificial intelligence putting words in our mouths.\u201d<\/p>\n Chan\u2019s campaign reportedly provided screenshots of the chatbot responding to questions about her accent and citizenship. In one response, the bot joked about Chan\u2019s San Francisco accent, while another made a reference to her citizenship status. According to the Chronicle<\/a>, Hong Kong-born Chan immigrated to the US from Taiwan at age 13 and is a non-native English speaker.<\/p>\n Wiener\u2019s campaign denied designing the chatbot to produce racist or sexist responses, saying it was political satire focused on Chan\u2019s record, accusing her of using race and gender to deflect criticism.<\/p>\n Wiener’s campaign spokesperson Joe Arellano said the chatbot was meant to lean into the \u201cAI craze overtaking the city\u201d while educating voters about Chan\u2019s record.<\/p>\n The Chronicle reports that Chan’s spokesperson Ian Krager accused Wiener of refusing to take responsibility for the chatbot\u2019s responses.<\/p>\n \u201cInstead of apologizing or taking responsibility for his bigotry bot, Senator Wiener is blaming everyone else,\u201d Krager said.<\/p>\n The dispute has also raised questions about how effectively AI chatbots can be prevented from reproducing stereotypes. Valentin Hofmann, an assistant professor at the University of Munich who has researched racial prejudice in AI models, said chatbots can draw on stereotypes embedded in the vast amount of internet data used to train them.<\/p>\n Hofmann told the Chronicle that accent-related stereotypes can be particularly difficult to remove because they can be encoded in subtle ways. After reviewing the responses provided by Chan\u2019s campaign, he said the chatbot appeared to have picked up information about Chan\u2019s background and used her accent in a failed attempt at satire.<\/p>\n By the end of the week, the chatbot was reportedly no longer producing the responses that sparked the controversy. When asked about Chan\u2019s accent or citizenship, it instead said it did not have information on those subjects.<\/p>\n The controversy has also brought the candidates\u2019 competing AI policies into focus. Chan recently released a video outlining proposals that include giving workers affected by AI a greater role in decisions about its use, requiring human oversight in areas such as finance and transportation, and requiring AI companies to cover the full energy and infrastructure costs associated with data centers.<\/p>\n As the Chronicle reports, Chan has also said she supports a nationwide moratorium on new data center construction.<\/p>\n Wiener has pointed to his own record on AI policy while defending the chatbot, including his unsuccessful 2024 effort to pass Senate Bill 1047<\/a>, an AI safety bill opposed by Pelosi and other Democratic leaders that was ultimately vetoed by Governor Gavin Newsom.<\/p>\n California Lieutenant Governor Eleni Kounalakis also reposted Pelosi\u2019s letter and called for the chatbot to be taken down, saying AI-generated impersonation has no place in politics.<\/p>\n Wiener ‘s campaign reportedly received backing from a Super PAC funded by Anthropic, whose Claude chatbot powered the website.<\/p>\n