Garry Tan, president and CEO of Y Combinator, mentioned at The Economic Club of Washington, D.C. this week that “regulation is likely necessary” for artificial intelligence.
In a one-on-one interview with Teresa Carlson, a General Catalyst board member, Tan discussed various topics including how to get into Y Combinator and his views on AI. He emphasized that “there is no better time to be working in technology than right now.”
Tan expressed his “overall supportive” stance on the National Institute of Standards and Technology (NIST)’s effort to develop a GenAI risk mitigation framework. He also noted that “large parts of the executive order by the Biden Administration are probably on the right track.”
NIST’s framework suggests that GenAI should comply with existing laws on data privacy and copyright, disclose its use to end users, and establish regulations to ban it from creating harmful materials. Biden’s executive order adds further measures like requiring AI companies to share safety data with the government and ensuring small developers have fair access.
However, Tan, like many Silicon Valley VCs, expressed reservations about other regulatory efforts, labeling bills related to AI in California and San Francisco legislatures as “very concerning.”
One contentious California bill, mentioned by Politico, is proposed by state Sen. Scott Wiener. It would allow the attorney general to sue AI companies if their products cause harm.
“The big discussion broadly in terms of policy right now is what does a good version of this really look like?” Tan noted. He highlighted that figures like Ian Hogarth in the UK are being thoughtful about balancing power concentration and innovation while mitigating potential harms.
Hogarth, a former YC entrepreneur and AI expert, is involved in the UK’s AI model taskforce.
“The thing that scares me is that if we try to address a sci-fi concern that is not present at hand,” Tan said.
Regarding Y Combinator’s approach to responsibility, Tan explained that if YC disagrees with a startup’s mission or the societal impact of its product, “YC just doesn’t fund it.” He mentioned there were several instances when startups reported in the media had been rejected by YC for not aligning with their values.
“We go back and look at the interview notes, and it’s like, we don’t think this is good for society. And thankfully, we didn’t fund it,” he added.
Artificial intelligence leaders keep messing up
Tan’s policy still allows Y Combinator to produce many AI startups. As reported by Kyle Wiggers, the Winter 2024 cohort included 86 AI startups, nearly double from Winter 2023 and almost triple from Winter 2021, per YC’s official startup directory.
Recent events cast doubt on whether AI companies can be trusted to define responsible AI. Truth Voices reported that OpenAI is disbanding its AI responsibility team.
There was also a controversy where OpenAI used a voice resembling actress Scarlett Johansson’s in a demo for its new GPT-4o model without her approval. OpenAI removed the voice after denial, creating questions about Sam Altman’s ethics, as discussed on Where’s Your Ed.
Meta created its own AI news by announcing an AI advisory council composed solely of white men, overlooking women and people of color who have significantly contributed to AI development and innovation.
While Tan did not mention these incidents, like most Silicon Valley VCs, he sees opportunities for large, profitable businesses.
“We like to think about startups as an idea maze,” Tan explained. “With new technologies like large language models, the entire idea maze gets reshuffled. ChatGPT itself was one of the fastest consumer products to succeed in recent memory, which is good news for founders.”
Artificial intelligence of the future
Tan also said that San Francisco is at the heart of the AI movement. For instance, Anthropic, and OpenAI, both have roots in the city.
Tan joked that he won’t follow in Sam Altman’s footsteps, remarking, “Altman had my job a number of years ago, so no plans on starting an AI lab.”
One of YC’s success stories is legal tech startup Casetext, which sold to Thomson Reuters for $600 million in 2023. Tan noted Casetext was one of the first firms worldwide to access generative AI, leading to one of the first exits in generative AI.
Looking ahead, Tan asserted the need for a smart approach to AI, given potential risks like bioterror and cyber attacks, advocating for “a much more measured approach.”
He also believes there won’t be a “winner take all” situation, but rather a diverse marketplace offering consumer choices and freedom for founders to create something impactful.
Tan’s primary concern isn’t the fear of rogue AIs but the lack of variety in AI options.
“We might actually find ourselves in this other really monopolistic situation where there’s great concentration in just a few models. Then you’re talking about rent extraction, and you have a world that I don’t want to live in.”
