Garry Tan, president and CEO of Y Combinator, recently addressed an audience at The Economic Club of Washington, D.C., where he emphasized that “regulation is likely necessary” for the rapidly evolving field of artificial intelligence (AI). In a one-on-one interview with Teresa Carlson, a board member at General Catalyst, Tan tackled various topics, including the pathways to joining Y Combinator and the current technological landscape. He confidently stated, “There is no better time to be working in technology than right now.”
Tan expressed strong support for the National Institute of Standards and Technology (NIST) in its efforts to create a comprehensive risk mitigation framework for generative AI. He noted that “large aspects of the Biden administration's executive order (EO) appear to be on the right path.” NIST’s proposed framework includes essential measures, such as ensuring that generative AI adheres to existing laws governing data privacy and copyright, mandating the disclosure of AI usage to end users, and implementing regulations to prohibit AI from creating harmful content like child sexual abuse materials. The executive order consists of various directives aimed at promoting safety in AI, ensuring small developers have equitable access, and obligating AI companies to share safety data with the government.
However, like many venture capitalists in Silicon Valley, Tan expressed concerns about certain regulatory initiatives. He highlighted bills in California and San Francisco that he finds “very concerning.” One notable proposal from state Sen. Scott Wiener would empower the attorney general to sue AI companies for any harmful effects their products may cause, as reported by Politico.
“The critical policy discussion now revolves around what a responsible regulatory approach looks like,” Tan explained. He cited Ian Hogarth in the U.K., who has been charged with creating an AI model taskforce. “Hogarth’s perspective emphasizes the importance of considering the concentration of power while simultaneously promoting innovation and minimizing potential harms.”
Tan elaborated on how Y Combinator manages its responsibilities regarding startup funding. He noted that if the organization disagrees with a startup’s mission or its societal impact, “YC just doesn’t fund it.” He recounted instances when a company that drew media attention ended up not receiving funding after the team reviewed their application notes.
Despite these guidelines, Y Combinator continues to cultivate a significant number of AI startups. The Winter 2024 cohort featured 86 AI startups, nearly double the amount from Winter 2023 and close to triple the number from Winter 2021, according to Y Combinator’s official startup directory. Recent controversies around AI companies have raised questions about the trustworthiness of those developing AI products. For example, tech media reported that OpenAI disbanded its AI responsibility team, and controversy arose when the company used a voice that resembled actress Scarlett Johansson’s without her consent during a demonstration of its new GPT-4o model. OpenAI later retracted the voice but denied any intent to imitate Johansson. Concerns around OpenAI’s management practices also surfaced, particularly regarding its handling of employee equity.
Meanwhile, Meta faced backlash for announcing an AI advisory council composed entirely of white men, effectively excluding women and people of color who significantly contributed to AI's creation and growth. Although Tan did not mention these instances directly, he, like many Silicon Valley investors, views them as opportunities for growth and innovation.
“We envision startups as an idea maze,” Tan remarked. “As new technologies like large language models emerge, the landscape shifts dramatically. ChatGPT emerged as one of the fastest consumer products to achieve success in recent history, which bodes well for founders.”
Tan noted that the heart of the AI movement resides in San Francisco, home to startup Anthropic, initiated by Y Combinator alumni, and OpenAI, a Y Combinator spinoff. Tan humorously remarked that he has no plans to follow in Sam Altman’s footsteps, considering Altman previously held his position.
Referring to Y Combinator's success story Casetext, which was acquired by Thomson Reuters for $600 million in 2023, Tan recognized it as one of the pioneers in harnessing generative AI for legal tech. Looking ahead, he emphasized the need for a cautious approach toward AI, particularly regarding risks associated with bioterrorism and cyberattacks. “We must adopt a much more measured approach to this technology,” he asserted.
Tan believes there won’t be a single dominant player in the AI field but instead a diverse range of options that empower founders to create impactful solutions that can reach billions. Notably, he fears that we could end up in a monopolistic situation if there are only a few dominant models, leading to problematic market behaviors.
“What troubles me most is not malevolent AI but a lack of diversity in AI offerings,” Tan revealed. “We may find ourselves in an environment characterized by a concentrated monopoly, which would result in undesirable economic consequences.”
Stay tuned for more AI insights as we launch our dedicated AI newsletter, coming soon.
Ranking
