The AI safety debate is tearing Silicon Valley apart

It’s not just OpenAI.

The long-simmering fault lines within OpenAI over questions of safety with regard to the deployment of large language models like GPT, the engine behind OpenAI’s ChatGPT and DALL-E services, came to a head on Friday when the organization’s nonprofit board of directors voted to fire then-CEO Sam Altman. In a brief blog post, the board said that Altman had not been “consistently candid in his communications.” Now rumors are swirling about Altman’s next move—and possible return.

But OpenAI is not the only place in Silicon Valley where skirmishes over AI safety have exploded into all-out war. On Twitter, there are two camps: the safety-first technocrats, led by venture firms like General Catalyst in partnership with the White House; and the self-described “techno-optimists,” led by libertarian-leaning firms like Andreessen Horowitz.

The technocrats are making safety commitments and forming committees and establishing nonprofits. They recognize AI’s power and they believe that the best way to harness it is through cross-disciplinary collaboration.

Hemant Taneja, CEO and managing director of General Catalyst, announced on Tuesday that he had led more than 35 venture capital firms and 15 companies to sign a set of “Responsible AI” commitments authored by Responsible Innovation Labs, a nonprofit he cofounded. The group also published a 15-page Responsible AI Protocol, which Taneja described on X as a “practical how-to playbook.”

Taneja’s tweet was quickly ratioed. Praying for Exits, a Silicon Valley meme account and investor, posted a screenshot of messages between an AI researcher, named Rohan Pandey, and an investor at Insight Partners, which also signed the Responsible AI commitments, in which Pandey canceled their upcoming meeting; Pandey said the commitments would “endanger open-source AI research & contribute to regulatory capture.”

Meanwhile, the techno-optimists (some refer to themselves as “humanists” and their adversaries as “doomers”) are grandstanding via tweet, podcast, and blog post. They recognize AI’s power, and they believe that entrepreneurs—rather than policymakers—are best equipped to make the technology a force for good.

“I want to build an AI agent that I am okay with giving to my children,” AI researcher and cognitive scientist Joscha Bach tweeted on Thursday. “It’s the same AI I want to build for myself. It will talk to me, join me, extend me. It will fit like a glove around my mind, and you will want one, too.”

Venture capitalist Marc Andreessen replied, “100 emoji.”

Notably, the techno-optimists have been up in arms since President Biden issued an executive order on the “Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence” in late October. In Europe, negotiations over an AI Act remain underway, with debate centering around the question of regulating foundation models directly as opposed to regulating consumer harm.

“Make no mistake, policymakers view the AI battle as an opportunity to relitigate tech regulation more broadly,” Andreessen Horowitz general partner Martin Casado tweeted earlier this week. “It is absolutely imperative we don’t let this happen.”

As news broke of Altman’s ouster, and OpenAI board chairman Greg Brockman’s subsequent resignation, techno-optimist Twitter wasted no time in assigning hero and villain labels to the players involved.

“Unsure what the board saw, from the outside looking in, it’s hard to argue there’s a better AI Exec than Sam,” Two Sigma investor Vin Sachidananda tweeted. “Also, given the inherent challenges of developing these technologies seems strange that half the board consists of ‘AI governance’ folks that have never built things.”

As for where Altman stands in the debate, he has done plenty of artificial general intelligence (AGI) fear-mongering, a techno-optimist no-no. But at the same time, he has pushed to deploy and commercialize AI rapidly, taking a technology that was once the purview of the ivory tower into the mainstream—a techno-optimist victory. Plus, he’s now the victim of bureaucracy (albeit one that he himself created).

Regardless of what Altman says or does, as the bellwether of AI, each side will claim him as their own—and the debate will continue.

https://www.fastcompany.com/90985788/openai-ai-safety-debate-silicon-valley-andreessen-general-catalyst?partner=rss&utm_source=rss&utm_medium=feed&utm_campaign=rss+fastcompany&utm_content=rss