The Probability Paradox: Why Actuaries Can't Price AI
When the masters of risk management meet the black box, the only winning move is not to play.
If you want to see where the AI hype cycle is actually starting to wobble, don’t stare at the VC crowd. Look at the actuaries. They’re quietly doing something way more telling.
I was reading a Financial Times piece today, and tucked between all those corporate-scented paragraphs was a surprisingly bold move: big insurers—AIG, Great American, WR Berkley—are going to U.S. regulators asking for permission to rewrite their policies so they can explicitly exclude AI-related risks. Mostly the generative stuff: LLMs, chatbots, the usual suspects.
On the surface, sure, you can shrug and call it routine risk management. But the more I sat with it, the more it felt like a snapshot of where we really are with AI—far past the glossy keynote slides and deep into the “wait, hold on” phase.
What really jumped out was the weird philosophical contrast. Actuaries are basically the OG probability people. Their whole profession is built on taming uncertainty with math and mountains of historical data.
And now they’re looking at the ultimate probability engines (LLM) and saying, “Nope. Too weird. Too opaque. Not touching that.”
Dennis Bertram from Mosaic Insurance says as much in the FT article. He calls these models a “black box,” and he’s not being metaphorical. He’s saying: we don’t know what’s going on inside, and that’s a problem.
This isn’t just about hallucinations or some chatbot inventing a fictional court case. The real issue is structural.
Insurance works because disasters are usually isolated. One company’s outage doesn’t automatically take down everyone else. Insurers love that kind of independence because it is predictable, clean, and insurable.
But AI throws a wrench in that. As Kevin Kalinich from Aon puts it (in the FT article), we’re now dealing with “systemic, correlated, aggregated risk.” Let me translate: if everyone is using the same foundational model, one glitch doesn’t produce one claim; it produces a flood.
Insurers can stomach a $500 million loss when someone screws up individually. What they can’t price—let alone survive—is a model error that generates a thousand losses all at once. Or ten thousand.
So when AIG says AI risks will “likely increase over time” and then immediately goes for blanket exclusions, that’s not subtle. They’re saying “no thanks, we’re out.” rather than “it’s risky but manageable.”
And that hits differently.
Because once the actuaries look at shiny new AI revolution and decide it’s a bet they won’t underwrite, that tells something: the actual, lived risk structure forming underneath all the hype.
About Me
My name is Jonathan Chen. My career has been defined by a rare privilege: seeing China’s most significant stories from both sides.
First, as the outsider breaking the news. Then, as the insider managing the narrative.
As an investigative reporter for China’s most respected outlets, including Southern Metropolis Daily and the 21st Century Business Herald, my job was to uncover the truth. I was the first reporter in China to disclose the Neil Heywood poisoning case and uncovered other major scandals that influenced the country’s political landscape, earning multiple news awards. My work began after an internship at The New York Times‘ Shanghai bureau.
Then, I moved inside.
For over a decade, I led brand and public relations for major corporations, including real estate giants like Vanke and, most recently, as the Head of Public Relations for the Hong Kong-listed gaming company.
I learned how China’s real estate, internet, and AI industries actually operate—from the inside. I wasn’t just observing; I was in the room.
Why Subscribe?
This Substack is where my two worlds collide.
I am not a ‘China watcher’ observing from afar. I am not a rumor-monger. I am an analyst with a source list built over two decades—first as a reporter, now as an industry principal.
When you subscribe, you get high-signal, low-noise analysis you can’t find anywhere else:
Deep Dives into China’s political-economic trends, internet regulation, and the AI industry.
Insights, Not Just News: I connect the dots between public policy, corporate strategy, and market movements, based on verifiable information and executive-level sourcing.
The “How it Works” Perspective: Understanding the hidden logic of China’s property sector, the games industry, and more.
My analysis is frequently sought by major outlets like Bloomberg, Nikkei Asia, and Reuters, and I’m a regular guest on The Paper (澎湃新闻).
If you are tired of the noise and want credible, high-fidelity insights, you’ve found the right place.