And the dice may already have been rolled…
A stranger knocks on your front door this afternoon.
“Got a proposition for you,” they say when you open.
“I’ve got three coins in my pocket. Pick one.
“Two of them mean your kids and grandkids will live lives about 55 times richer than yours. They’ll enjoy technologies you can’t even fathom. Diseases that plague us today will largely be history. Living standards that make today’s world look positively primitive.
“And the third coin…?” you ask.
“The human race goes extinct.”
Would you take that bet?
I’m betting the vast majority of people would say “hell no” and slam the door shut.
But apparently, at least one very smart economist thinks you should pick a coin…
This is a warning: The next paragraph you read is going to bring on narcolepsy. But power through, because the rest of today’s dispatch is much more interesting.
“What is the price of this amazing change in living standards? Recall that we would face a flow probability of existential risk of 1% per year for 40 years, so the probability we survive this A.I. explosion is exp(−.01 × 40) ≈ 0.67. In other words, with log utility it is optimal to take a 1 in 3 chance of ending human existence in exchange for a 2/3 chance of dramatically raising living standards by a factor of 55.”
Wake up!
That collection of what I think are words is gobbledy-gibberish for 107% of humanity.
What I can tell you is that it comes from our aforementioned very smart economist, a guy named Chad Jones, who the Financial Times described thusly: “a brainy, prominent economics professor at Stanford, where he has specialized in the impact of technology on growth.”
Apparently, The Chad has been hired by Anthropic, the company behind Claude AI, which I use all the time in research (honestly, it’s brilliant, but not without its shortcomings). The quote above, which I’m still not convinced is a real form of English, came out of a 2023 paper titled “AI dilemma: Growth vs. Existential Risk.”
To be fully transparent with you, I went to Claude and input that quote and asked for a translation in terms a human brain can understand. The nut of it is this: If you run the numbers the way economists run numbers on any investment, the AI bet pencils out because there is a 67% chance of a 55x leap in living standards. Not 55%… 55 times, or basically the entirety of the Industrial Revolution compressed into a random Tuesday.
The downside: a 33% chance that AI makes humans extinct inside 40 years.
Present that data to an economist but leave out the context of humanity’s expiration date, and the answer is, “Yep, that’s a fair tradeoff. A two-thirds chance to 55x your living standard? Sign me up!”
But then throw in the “oh by the way, there’s a one in three chance that you wake up and die somewhere along the road over the next 40 years.”
Thing is, we kinda don’t have the option to reject the wager.
AI is here and it’s growing smarter every day. It’s being more deeply integrated into lives and systems every day. Do we reach a point where AI is in control of medical decisions and runs the numbers on a life-saving surgery but determines by way of a mechanically cold and heartless calculation that offering this particular treatment to this particular person isn’t worth the cost to society at large… next, please.
Maybe this can be stopped. Maybe not. Hard to say—largely because I don’t trust the men and women propelling AI. They’re largely billionaires and their actions seem to imply that their primary interest is pocketing their next billion dollars, regardless of the impact on humanity.
And look, I get it—I’m part of the problem. I use AI multiple times a day in my research because it makes me vastly more efficient. So the $20 I’m paying monthly for my Claude subscription is 20 bucks that goes toward the extinction of my own species.
But here’s the bright light at the end of the tunnel that cold economic calculations rarely account for: Across every transformative technology in human history, the people who control it first are never the people who end up answerable to it.
Nuclear weapons were built by a handful of physicists and generals who could’ve ended the world on a Tuesday afternoon with nobody’s permission. But after Hiroshima, humans stepped in and we got three decades of treaties, inspections, and a public that refused to let the people with the button operate unsupervised.
That wasn’t because the people holding the button had a change of heart. Rather, it was because the rest of us forced the issue.
Or, heck, let’s go back much deeper into the past: the printing press.
Gutenberg didn’t roll it out with a content moderation policy. That contraption is directly responsible for a century of religious wars before Europe figured out how to live with mass information. Messy, violent, expensive — but the humans figured it out.
AI right now is in its unaccountable-billionaire phase.
That’s not a permanent condition, however. It’s just the starting condition.
The Chad Joneses of the world can keep running their cold little equations, and the Sam Altmans can keep chasing their next billion — but they don’t actually get to decide alone how this plays out.
Humans, collectively, will do that by demanding guardrails, by refusing to let “it’s just math” be the final word on whether Granny gets her gallbladder surgery.
The extinction math is real.
But so is the leverage the rest of us still have to make sure it never gets tested.
Not signed up to Jeff’s Field Notes?
Sign up for FREE by entering your email in the box below and you’ll get his latest insights and analysis delivered direct to your inbox every day (you can unsubscribe at any time). Plus, when you sign up now, you’ll receive a FREE report and bonus video on how to get a second passport. Simply enter your email below to get started.
