Illustration: Maura Losch/Axios
A new partnership between a tech nonprofit, leading venture capital firms and the Department of Commerce aims for startups to bake responsibility into their AI applications as they grow.
Driving the news: Responsible Innovation Labs, which works with startup founders and investors, launched the Voluntary Responsible AI Protocol on Tuesday alongside Commerce Secretary Gina Raimondo in San Francisco.
- The protocol focuses on organizational buy-in on responsible AI; transparency, risk and benefit forecasting for AI; auditing and testing; and ongoing improvements.
- It lays out tools for startups to do testing and best practices following guidelines made with academics, civil-society groups and Commerce.
- Thirty-five investors signed on to the effort, including Lux Capital, General Catalyst, Bain Capital and Mayfield.
What they're saying: The partnership is "what we think is the first set of industry-driven responsible AI commitments and protocols really built for the startup and early-stage companies and investors who back them," Gaurab Bansal, executive director of RIL, told Axios.
- There are plenty of ways for larger, more well-resourced companies to adopt responsible innovation for AI, but they're not practical for early-stage startups, he said.
- Bansal said RIL spoke with Biden administration officials a number of times before releasing the protocol, and the group is spending more time on Capitol Hill.
Details: The commitments are voluntary, he said, but the point is "trying to bring about a mindset of responsibility, making it part of the operational norms of company-building."
- "Hopefully this builds a movement of operators and investors who say 'We want to lead on this, even if regulators may or may not act in the near term,'" Bansal said.
State of play: People are apprehensive about the potential of AI. Venture capitalists need AI companies to gain users and make money, and some VCs say responsible innovation protocols will help them fund more successful startups.
- "We're investing in these companies building AI technology, and we want it to be adopted broadly," Shahin Farshchi, general partner at Lux Capital, one of the participating VCs, told Axios.
- "The way that happens is there is a sense of comfort, where there is assurance, where folks are responsible for compliance.… The way to do that is to be engaged with the government and the regulators."
- Bansal said: "We do believe trust is going to actually lead to scale, particularly in AI.… We think there is a business case for why building responsibility matters, and we think it's more efficient to build it on the front end."
The bottom line: "People know that this will become widespread," Farshchi said, comparing the explosion of AI to the advent of social media, which he said was not expected to as be widely adopted and did not put safety at the forefront.
- "So they are getting in front of all these potential issues that could happen as a result of it being widespread."
