During a recent event at The Economic Club of Washington, D.C., Garry Tan, the president and CEO of Y Combinator, expressed his belief that artificial intelligence may require regulatory oversight.
In a conversation with Teresa Carlson, a board member of General Catalyst, Tan covered a variety of topics, including entry into Y Combinator and the current state of technology, highlighting the present as an unparalleled time for tech engagement.
Tan showed general agreement with the National Institute of Standards and Technology’s (NIST) efforts to create a framework for mitigating risks associated with generative AI (GenAI), and he believes that significant aspects of the Executive Order by the Biden Administration are likely on the correct path.
The NIST framework suggests measures such as ensuring GenAI’s adherence to existing legal standards concerning data privacy and copyright, making GenAI’s usage transparent to users, and formulating rules to prevent GenAI from generating content related to child sexual abuse. The Executive Order by President Biden encompasses a broad spectrum of mandates, including the obligation for AI firms to divulge safety data to the government and the assurance of equitable opportunities for smaller developers.
However, Tan shares the concerns of many Silicon Valley venture capitalists regarding certain legislative proposals concerning AI in California and San Francisco, describing them as “very concerning.”
A particular bill introduced by California State Senator Scott Wiener is currently the subject of much debate. This proposed legislation would empower the state’s attorney general to initiate legal action against AI companies for producing harmful products, as reported by Politico.
Garry Tan highlighted the ongoing, wide-ranging policy debate, questioning the ideal form such regulations should take. He pointed to UK-based Ian Hogarth as a source of considered insights. Hogarth is cognizant of the potential for power consolidation and is actively exploring ways to foster technological innovation while simultaneously preventing the most severe risks associated with it.
Ian Hogarth, an AI specialist and former entrepreneur with Y Combinator (YC), has been selected by the UK government to serve on a taskforce dedicated to AI models.
Garry Tan, president of YC, voiced his concerns about preemptively tackling hypothetical, science fiction-like issues that are not currently pressing.
Regarding YC’s ethical stance, Tan emphasized that the incubator refrains from funding startups whose missions or potential societal impact do not align with YC’s values. He recalled instances where, upon encountering media coverage of certain companies that had applied to YC, the team would review their interview notes and often conclude that the company’s direction was not beneficial for society, leading to a decision not to provide funding.
Artificial intelligence leaders keep messing up
Garry Tan’s approach allows Y Combinator to continue producing a significant number of AI startups from its cohorts. According to a report by Kyle Wiggers, the Winter 2024 cohort featured 86 AI startups, almost twice the number from Winter 2023 and nearly three times that of Winter 2021, as per YC’s official startup directory.
Recent developments have raised questions about the credibility of AI product vendors in setting standards for responsible AI. For instance, TechCrunch revealed last week that OpenAI is disbanding its AI ethics team.
Another controversy arose when OpenAI used a voice resembling that of actress Scarlett Johansson for the demonstration of its new GPT-4o model. Although Johansson declined the use of her voice, OpenAI later removed the ‘Sky’ voice, denying any imitation of Johansson. This incident, along with concerns about OpenAI’s policies on employee equity, has led to public scrutiny of Sam Altman’s ethical judgment.
In another event, Meta announced the formation of an AI advisory council composed solely of white men, excluding women and people of color who have been instrumental in the AI field.
Tan did not comment on these specific situations. Reflecting the typical Silicon Valley VC mindset, he focuses on the potential for creating substantial and profitable new enterprises.
Tan described startups as navigating through an “idea maze,” where emerging technologies like large language models disrupt the existing landscape. He cited ChatGPT as one of the most rapidly successful consumer products in recent times, which he views as a positive development for entrepreneurs.
Artificial intelligence of the future
Garry Tan highlighted San Francisco’s pivotal role in the AI revolution, citing Anthropic and OpenAI as notable examples that originated there, both with ties to Y Combinator (YC).
Tan humorously remarked that he has no intention of emulating Sam Altman’s career trajectory by establishing an AI lab, despite Altman having previously held Tan’s position at YC.
Another YC-affiliated success story is Casetext, a legal technology firm that was acquired by Thomson Reuters for $600 million in 2023. Tan regards Casetext as one of the initial companies to leverage generative AI, leading to one of the earliest lucrative exits in the field.
Looking ahead, Tan emphasizes the necessity for cautious and intelligent engagement with AI technology, particularly considering the potential threats of bioterrorism and cyber attacks. He advocates for a balanced approach to AI development.
Tan envisions a diverse AI ecosystem, not dominated by a single entity, but rather a flourishing landscape offering a multitude of choices for consumers and opportunities for entrepreneurs to impact billions.
His ideal scenario is a thriving startup environment that benefits both YC and its investors. Tan’s primary concern isn’t the fear of uncontrollable AI, but rather the prospect of limited AI options leading to a monopolistic market dominated by a few models, which he believes would be detrimental to the industry.