Programming Note: This newsletter was previously called Steady State. It's now called KCRW Context, and it's still written by KCRW's Editor-in-Chief Arnie Seipel.

AI leaders like OpenAI CEO Sam Altman, seen above at the White House last year, have developed close relationships with the White House. Though, as those top AI figures warn of the dangers the technology could bring, President Trump has dismissed the warnings as a “hoax.” Photo by Andrew Harnik/Getty Images.

Hi there -

It’s impossible to ignore the warnings about AI that have ramped up over the last week. Notably, an Anthropic researcher resigned and stated publicly that there is a greater than 10% chance artificial intelligence could wipe out the human race.

But how?

There are several scenarios that have been imagined, but one of the clearest isn’t that far from reality. We know that foreign powers like Iran have attempted to hack into U.S. water and electricity infrastructure. Imagine AI systems getting to the point where a nefarious actor could use them to cause widespread destruction to systems that deliver essential resources. Or perhaps worse, imagine if the AI system takes it upon itself to do the hacking.

That’s just one possible nightmare.

Sarah Isgur from KCRW’s Left, Right & Center offered an apt metaphor on Friday as we marked the 25th anniversary of the September 11th attacks. “We had warnings before 9/11. In that sense, it feels very similar to this AI moment,” she said. “We have warnings that this is an important issue that we need to focus on, but we're not focused on it.”

That focus does seem to be shifting.

The warnings from a researcher weren’t dismissed by AI leaders. They were amplified.

Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and even Elon Musk all called for a slowdown in AI development in the past week. Skepticism abounds, since there is a history of tech companies shifting the burden of responsibility to government regulation, and then undermining those efforts.

So what do we do about it?

This is an international problem, not something that the U.S. can deal with just within its own borders. One framework being discussed a lot is to think of AI like nuclear material. It needs to be regulated as we seek out its beneficial uses, just like nuclear power, and tightly restricted so it cannot be used for destruction, as we do with nuclear weapons.

Nuclear nonproliferation required an improbable partnership with the U.S. and the Soviet Union. Keeping AI from getting out of control will require cooperation between the U.S. and China. The leaders of both countries have been doing the opposite — approaching this as an arms race, akin to the Cold War.

Trump frames the issue of AI development as a zero-sum, existential battle with China. In the last few days, he’s called the warnings a “hoax” and said that his leadership is all that’s needed to keep it under control, not any formal regulation.

Just as there are members of Trump’s own party sounding the alarm while he downplays things, the head of China’s intelligence agency warned that AI poses a threat to the stability of his country.

China’s leader Xi Jinping has pushed back on the warnings, suggesting that Beijing could lead the development of international AI governance among nations in the global south, essentially a technological sphere of influence that Xi could establish in competition with the U.S.

It’s yet another parallel to the Cold War, but these similarities begin to break down when facing the pace of artificial intelligence growth and the nature of conflict around the world today.

The development is far faster than doomsday bombs, and AI’s capabilities are possibly beyond what humans can conceive of.

Its accessibility isn’t limited to only global superpowers, removing the threat of mutually-assured destruction that kept the Americans and Soviets out of nuclear war.

Launching an AI-powered weapon requires an internet connection, not an intercontinental ballistic missile. Since proximity is not a factor, the need for international cooperation before the nightmare scenario plays out is apparent.

Nuclear nonproliferation is a global effort that took decades to yield its first real progress. With AI, we just don't have that kind of time.

GO DEEPER

🧑‍💻 A theory that AI could be president may sound ridiculous, but the Left, Right and Center crew talked through how the presidency could be driven by AI decision-making in the future. Plus, they dive into Trump’s strategy to make the midterms all about himself. Listen here.

🇷🇺 ProPublica: “Donald Trump Jr.’s Bahamas wedding was secretly bankrolled by Russian oligarch close to Putin”

🔴 Roy Cohn, the sidekick of Sen. Joe McCarthy who helped orchestrate the Red Scare, shaped Donald Trump to a degree almost beyond belief. Historian Kai Bird wrote a new biography of Cohn that connects the dots. His conversation with KCRW’s Press Play starts around the 25:00 mark here.

Take care,
Arnie Seipel
KCRW Editor-in-Chief
 

If someone forwarded this to you and you want to subscribe, click here.