Can We Trust AI Companies?
Artificial intelligence is certainly helpful. Who by now hasn’t used, or at least knows someone who has used, ChatGPT or another large language model to help with personal or business tasks?
But it’s increasingly clear that AI poses risks with weekly, if not daily, reminders. Take Google’s clumsy AI Overview search summaries. On a humorous level, they have told people to eat rocks and (non-toxic) glue. On a troubling level, many in the media feel it will suck the web traffic, and thus the advertising lifeblood, out of independent journalism. And the dangers don’t stop there.
?At the recent Milken Global Conference, Worth spoke with Robert Silvers, the undersecretary in the Office of Policy at the U.S. Department of Homeland Security, about a broader set of risks. These include the use of AI in mission-critical sectors such as hospitals, power plants, water-treatment facilities, and air transportation—where errors can be much more serious than bonkers culinary suggestions. AI can also supercharge cyber attacks on these facilities and sectors.?
To address these and other dangers, DHS is creating an AI Safety and Security Advisory Board consisting of government officials, civil society representatives, and industry leaders such as OpenAI CEO Sam Altman and Google CEO Sundar Pichai. This represents the American approach to AI safety.?
“We are right now working in the domain of voluntary commitments with some of the frontier model developers in the AI world,” said Silvers. “We don’t want to squash the innovation. We want to study it closely. We want to work together with industry to understand it so we can do things smartly.” (However, he concedes that some uses of AI will likely be regulated in the future.)
Without a legal authority to regulate AI, as the EU has, the U.S. depends on the integrity and goodwill of the industry to do the right thing. Can we trust the heads of AI companies in this? Not according to Sam Altman. When asked if people could trust him, he said, “You shouldn't. If this really works, it’s like quite a powerful technology. You should not trust one company and certainly not one person with that."?
That quote is featured in "The OpenAI Story," the latest season of Bloomberg's Foundering podcast series. It probes deeply into Altman’s character as it’s developed from childhood to the present. Give it a listen. (Unlike most Bloomberg content, it’s not behind a paywall.) If we put our faith in AI leaders to do the right thing, we need to know if that’s in their nature.
Worth will probe such issues around safety and trustworthiness of AI and other technologies at the Techonomy 24: Leading with Intelligence conference in Lake Nona, Florida, from November 17th to 19th. Please join us.
—Seán Captain, executive editor
This Week on Techonomy
DHS undersecretary discusses the risks of AI and how the U.S. government is responding. (Video)
Research suggests that grass-fed meat has a higher carbon footprint. (Courtesy of Sentient Media)
How the Microsoft cofounder and his partner are dedicating their resources to philanthropy.
These seven different types of watches are key to any good watch collection.
领英推荐
Tony Pan of Modern Hydrogen and Josh Lehman of Relyion, Inc. discuss the prospects for this challenging potential green fuel. (Video)
These siblings and ex-OpenAI execs aim to build a more responsible and ethical artificial intelligence company.
The latest research and innovations in helping women contend with this health milestone. (Video)
Techonomy Events
This transformative event will galvanize leaders, innovators, and influencers from diverse sectors to explore and amplify sustainable practices that can drive significant environmental changes. Register.??
Worth's premier tech conference will focus on transformative technologies such as AI, biotech, robotics, and automation as well as their privacy and regulatory implications. Request an invite.
What We’re Reading and Listening To
Investments by OpenAI CEO Sam Altman are raising ethical questions. (Wall Street Journal “Tech News Briefing” podcast)
Volvo is first to develop a “passport” tracking environmental impacts of its electric vehicles. (Reuters)
Google is embarrassed regarding AI Overviews feature and leaked internal documents about how its search works. (New York Times “Hard Fork” podcast)