Citrini Research thoughts
Paper was referenced in the journal this morning, just finished reading through it.
Obviously an unlikely outcome but have been thinking about how our world will change from AI. Curious what others think about it.
bump
It is a fine thought experiment.
There is a 100% chance this won't actually happen in the real world.
I'd put this at 95+% chance this won't happen (never say never since you want to be open to every possibility)
But yes, it is a ridiculous post. It assumes that the world has little to no friction, and that AI carries no real job upside. Yes, headcount might grow slowly or even be flat for largest companies -- but if that thesis holds true, then how in the world is it not true that there are a lot more companies that are started? Smaller companies yes in headcount terms, but # of companies should be WAY higher
Likewise, there are a lot of well paying 'blue-collar' areas in healthcare that could absorb a portion of white collar folks. Shortages in HC are most acute in lower COL areas anyway, which means net living costs come down even with similar / modestly lower pay.
Tech bro cope coming in strong in the comments, pretending they haven't been babysitting droids for the past 6 months
I think the paper makes accurate points about a lot of industries being commoditized and that a lot of the new roles AI is creating can also be done by ai. Overall fits in with my existing view that ai will compress headcount in the financial services industry. However, I do not believe that the hypothetical reality discussed in the paper is probable.
He misunderstands that blue collar wages would actually rise if productivity in other parts of the economy grows. The view that a reduction in transaction costs which are pure efficiency is bad for anyone except those rent seeking for money and some short term disruption is also wrong. The only thing maybe true is if profit goes entirely to those who don’t consume then you can have some downward spiral but those people would save and invest it which would either lower interest rates, increase asset prices, and / or drive investment. The equation has to balance. Yes there will be disruption and winners and losers
The savings/investment loop works in aggregate but you’re dodging the distributional question. The productivity gains go to whoever owns the capital, the job losses land on people who largely don’t. Lower rates and higher asset prices help if you have assets. And in Citrini’s scenario that capital doesn’t recirculate through wages anyway, it just gets funneled straight back into more AI compute and accelerates the same loop.
And we basically ran this experiment already post-2000. Productivity has compounded for decades, median real wages were flat for most of it. The trickle didn’t trickle.
Also the equation theoretically balances eventually but the transition itself is the problem. It could be years. Hypothetically…
Post tax and transfer wages for lower quintiles have grown robustly look at the economist’s recent article on this. The government has redistributed the gains reasonably well. That’s the final thing citroni misses - realistic policy response
The most depressing part of the piece is that people actually think it's insightful.
It's is far too zero sum and also doesn't have a coherent model of a world with powerful AI. If AI becomes truly as capable, widespread, and cheap as it is in the author's scenario, wouldn't we also have access to that AI as well and our marginal product of labor will an order of magnitude higher than it is today?
There's a bunch of other things in the article that don't make much sense.
The guy is a moron hiding behind pretty-sounding literature. He misses a million things:
I could keep going on about this but there's no point. The article has deep flaws in nearly every argument he makes
His piece is specifically framed as a left tail worst scenario thought experiment not a real argument that it will happen why are you so mad at him lol
I'm annoyed because he's sitting on massive long positions in semis and short positions on software so he's deliberately pumping a fear narrative to manipulate markets & people. Then he goes off and lies about it 'I didn't think the market would be so negative on it' which is ridiculous since he's one of the most popular tech substacks by subs -- in this jittery market environment, you really expect anyone to buy that crap?
You want to pump an AI doomer narrative? Fine. But then pumping this narrative to literally manipulate the market is where I draw the line
Ut ut repellendus odit laborum. Eum alias quo consequatur similique consectetur. Necessitatibus enim et qui velit sint nihil maxime. Possimus dolorem quis rerum maxime magnam aperiam.
Repudiandae aperiam eos quo dolor inventore molestias placeat laudantium. Cum animi veniam similique. Quis et et explicabo dignissimos.
Est consectetur aspernatur aut quam quia eius. Est sed voluptatem deleniti sed nulla. Consequatur quis dolorum saepe esse consequatur alias. Libero quam laboriosam dolor quos dicta omnis. Dolores reiciendis aut voluptatibus sit sit provident qui. Possimus qui aut temporibus eos iure qui.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...