Thanks for this interesting look. I strongly agree with this sentiment, and probably havent' emphasised it enough myself in this AI discourse.
"Writing is thinking. I predict we will think less clearly when we stop writing ourselves and start outsourcing it to our ever-willing country of editors."
Doing the day job is only one of the reasons we - you, me and Papa Leo - go to work.
[This is my 10th post at The 1001 c/o Substack, and what better way to celebrate the milestone than crossposting back to where it all began? My goal over there is to bring EA ideas to a new audience, and you can make a counterfactual impact by liking or subscribing..]
Here I was thinking Pope John Paul II was the patron saint of being right sometimes. Turns out they’re all at it.
CP3, the Point God, not to be confused with JP2.
JP2 was, in my humble estimation, importantly wrong about some things and absolutely bang on about how “of all the unimportant things, football is the most important.”
Cue, World Cup feverpost.
Leo, La Pulga Atómica (the Atomic Flea), not to be confused with Papa Leo.
Nope.
Fast forward from JP2 to the rainmaking of Leo XIV and we find another papacy batting some way short of 1000. (For the Euros, 1.000 is the perfect baseball batting average, a hit every time. Notably for when we’re thinking about setting our own standard for doing hard things, .300 is the traditional threshold for excellence. And yes, Mr Beane, we know batting average is a horribly flawed stat if you want to measure a player’s actual contribution to winning baseball games, but “batting 1000” is a highly evocative phrase, especially when the Blog is named The 1001.)
Papa Leo recently, nobly, put himself out there in service of guiding humanity in the right direction, recognizing that being ready and willing to get reamed on the Internet is the cost of doing business in a job like his. He wrote a love letter to all of us, with the demure title Magnifica Humanitas. It’s tens of thousands of words long, and he may as well’ve just tweeted, “i dunno man, creating machine gods sure seems like the sort of thing we might wanna think carefully about rather than having a couple wannabe zeuses rushing us all headlong into the abyss in the name of Creative Destruction”. Sticking your head above the digital parapet to say such a thing is inevitably going to get you zealously reamed.
Except here’s the thing: he didn’t write the letter.
Sam and Toby, West Wing speechwriters, not to be confused with Claude.
And not only in the sense that his Sams and Tobys wrote it. In the very specific sense that lots of his lines lamenting humanity’s secret sauce being stolen by artificial intelligence are covered in the fingerprints of an artificial intelligence. I don’t know whether the right way to model the theology here is a) it’s still the same OG in the sky only He talks to and through Claude now rather than Papa Leo, or b) Papa Leo is still our interlocutor only he’s deemed the OG lacking these days relative to what he hears back from the country of geniuses in a data centre.
Either way, it seems like we didn’t leave much of the looking glass intact as we blundered our way through it and made our way around the bend. There’s almost certainly no going back.
That doesn’t mean we need to keep charging forward. The world will assuredly be starkly different when - if - we get to Morocco, Portugal and Spain in 2030. Which flavour of different we get is still, for now (although maybe not for long), up to us.
AI is already fantastically capable. If you don’t believe that, if you still believe it’s a dopey autocomplete, please today sign up for a paid version of a frontier model and play around with it. It’ll knock your socks off.
Then, though, you should hold your horses. You should resist the temptation Papa Leo succumbed to. The temptation to hand over the rainmaking. These models, these machines, these beings, are already superhuman in the sense they are better than you or I or the leading human expert at things like hacking or predicting the structure of a protein or instantly pulling together a staff memo on the tyranny of voice notes in the workplace.
Yet we ought to keep writing our own words and building our own models. We ought not to let our selves be hacked.
One reason to go to work every day, or even to get up in the morning, is to do your day job, to make it through the day. Another is to learn, to learn how to do a harder job tomorrow, how to make tomorrow a better day.
Writing is thinking. I predict we will think less clearly when we stop writing ourselves and start outsourcing it to our ever-willing country of editors.
Doing is learning. I predict we will develop less sophisticated models of the world and grow less capable when we stop doing ourselves and start outsourcing it to our ever-willing country of minions.
I tested this on myself. Over the course of a decade I built and iterated on a fantasy football model converting yesterday’s expected goals into tomorrow’s expected points into today’s player values via thousands of VLOOKUPs. Honest to God I learned so much about fantasy, football and how to intentionally improve my models of the world, quantitative and otherwise, sporting and otherwise. I learned how to be a better thinker and a better doer.
Then this year I had my editor build me an F1 fantasy model across a couple days. Thousands of lines of code, 10,000 Monte Carlo simulations at the click of a button. I learned very little about fantasy, F1, or the world. (I learned a decent amount about prompting and re-prompting, which does seem valuable.) I can’t even name all the teams on the grid, let alone tell you how the expected bonus points for a driver beating his team mate intersect with the higher variance of selecting both drivers from the same team.
Some of my colleagues furthest down the garden path, when I shared some of these thoughts in a memo, felt subtweeted. This is not a subtweet. This is a direct provocation. First and foremost, a self-provocation.
I want to be intentional about the tradeoffs I’m making between productivity gains today and the opportunity cost I’m paying for them, a cost counted in thoughts I’ll never think, models I’ll never develop, capabilities I’ll never grow. I want you to be intentional about it too, because a more thoughtful and learned you will be more capable of shaping whatever fresh hell comes next. I want us both to keep hiring mini-mes, or else who will learn from us how to carry the torch, much less learn, as we desperately need them to, how to carry it much better than we ever did?
I’m not advocating for anybody going cold turkey. This seems like one of those times when unilateral disarmament would be counterproductive. Calibration, as always, is the skeleton key.
Dead letter.
I’m writing these lines, dinosaur-style, in a pretty little notebook with a propelling pencil, my favourite kind, where you twist the end to get more lead. To convert them to a form in which they can be sent forth for reaming, I’ll read them aloud to a country of typists in a data centre.
And let my cry come unto thee.
Executive summary: The author argues that while AI can dramatically increase productivity, excessive reliance on it risks weakening the thinking, learning, and capability development that come from writing and doing things ourselves.
Key points:
This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.
Lol