Large Language Models, Small Labor Market Effects (nber.org) 17
The abstract of a study featured on NBER: We examine the labor market effects of AI chatbots using two large-scale adoption surveys (late 2023 and 2024) covering 11 exposed occupations (25,000 workers, 7,000 workplaces), linked to matched employer-employee data in Denmark.
AI chatbots are now widespread -- most employers encourage their use, many deploy in-house models, and training initiatives are common. These firm-led investments boost adoption, narrow demographic gaps in take-up, enhance workplace utility, and create new job tasks. Yet, despite substantial investments, economic impacts remain minimal. Using difference-in-differences and employer policies as quasi-experimental variation, we estimate precise zeros: AI chatbots have had no significant impact on earnings or recorded hours in any occupation, with confidence intervals ruling out effects larger than 1%. Modest productivity gains (average time savings of 3%), combined with weak wage pass-through, help explain these limited labor market effects. Our findings challenge narratives of imminent labor market transformation due to Generative AI.
AI chatbots are now widespread -- most employers encourage their use, many deploy in-house models, and training initiatives are common. These firm-led investments boost adoption, narrow demographic gaps in take-up, enhance workplace utility, and create new job tasks. Yet, despite substantial investments, economic impacts remain minimal. Using difference-in-differences and employer policies as quasi-experimental variation, we estimate precise zeros: AI chatbots have had no significant impact on earnings or recorded hours in any occupation, with confidence intervals ruling out effects larger than 1%. Modest productivity gains (average time savings of 3%), combined with weak wage pass-through, help explain these limited labor market effects. Our findings challenge narratives of imminent labor market transformation due to Generative AI.
Drawbacks of the study (Score:5, Interesting)
This study uses survey results from one to two years ago. First, survey results are simply data the reflects personal perceptions. Second, the employment impacts of LLMs are arguably just emerging, so the old surveys are not so insightful. Third, the key result is a 3% average productivity gain, but what would be much more useful would be the top 10% of productivity gains versus the bottom 10% and why there is a difference. The average number is not as useful without an understanding of the distribution.
Re:Drawbacks of the study (Score:5, Interesting)
Second, the employment impacts of LLMs are arguably just emerging, so the old surveys are not so insightful
Also, the impacts may be quite volatile. LLMs might result in huge job displacement - until the discovery of some flaw which can't be corrected for renders them unsuitable.
Then there's the problem of models feeding off of each other's hallucinations. If that's as bad as some have suggested, retraining the models to keep them 'pure' could cost a lot. That would make their adoption less universal than AI's cheerleaders expect.
News Flash (Score:5, Informative)
AI is over-hyped and under-performs
From a software development perspective, I've found it's good at troubleshooting problems, much like Google and StackOverflow, but completely shit at writing code that works.
Re:News Flash (Score:5, Insightful)
Re:News Flash (Score:4, Interesting)
My experience echoes yours. GitHub Copilot often writes reasonable-looking code, but then it will do stupid stuff like repeat code I already wrote, in a way that won't even compile. I find the tool useful mostly for languages that aren't my best, and arcane syntax such as is used to manipulate JSON or XML within SQL fields, or PowerShell or TerraForm scripts.
Re: News Flash (Score:3)
The gif of a black cat smashing at the keyboard comes to mind. "Look ma! I'm doing something! I'm working!"
No, you're not. You've just caused more mess and debugging required than if I wrote this code myself in a quarter of the time.
The study completely misses the point (Score:5, Interesting)
Remember CEOs and ctOS aren't smart they are lucky and well connected. Nepo babies mostly.
So there is tons of crap out there that's being called AI but is really just run-of-the-mill automation. Remember that stupid shirt about how go away or I'll replace you with a very small Perl script? It's like that.
Finally a recession, big one, is coming. The 7 trillion dollars in tax cuts being rammed through the Senate or going to shake the bond market triggering a pretty large recession and there's no stopping it at this point. It's clear that senators and members of the House fear being unseated in a primary by the billionaires demanding the tax cuts more than they fear losing their seat in the general.
So what's going to happen is your boss is going to fire you or he's going to fire your coworker. Then he's going to replace whoever gets fired with "AI". Meaning some form of generic automation.
If it works then you will get a modest pay cut because of the weak labor market or you'll be unemployed if you're the one that gets picked for layoffs but at least your workload won't increase.
And if the new automation doesn't work, well you're either going to be unemployed while somebody else is working 80 hours a week to make up for a broken AI or you're going to be employed working 80 hours a week making up for that broken ai.
Think of an AI as your useless coworker whose slack you constantly have to pick up. It'll be like that only the lazy coworker gets fired. You're still going to work twice the hours to make up for it or they will fire you and replace you with one of the many many unemployed people in the job market.
And remember good enough is always good enough and your boss is always looking for a way to commoditize your skill set so that you can be easily replaced. They even have a name for it it's called taylorism.
Look at me I'm explaining. And that means I'm losing.
And what do you think CEOs will do since (Score:2, Insightful)
They are all psychopaths like Musk without exceptions and they all to starve us out as quickly as possible. Maybe if it wasn't run by corrupt psychopaths. But there's about as much chance of that happening as there is of slashdot adding Unitranscode support.
I guess it's also possible they are just psychopaths that wants to chop people's hands off. You've got those people out there too. There's a growing problem with baby boomers who just want to kill somebody and they're old so they decide what the hell I'm
Re:The study completely misses the point (Score:4, Informative)
Just to make the coming recession that much worse, la Presidenta's alleged administration wants to use LLM AI throughout the government. You can guess the spiel, "this will make government more efficient, cut waste, fraud, and abuse we just know is there but have been unable to find; hence we need LLM to ferret it out....and kill the Deep State."
What they are creating is a Deep State where all the information in gov. databases (think SS, Medicare, election databases, etc.) is being siphoned off by the little Maggots in DOGE. To make it worse, it is being handed to that other S. African Nazi, Peter Thiel. Once their project is complete, they will be able to control the Fed. Gov. from the White House alone, no need to follow any of those pesky rules the Congress imposes to make gov. bow to the constituents.
The Dept. of Injustice is already asking for voter information, a lot of it, starting with Colorado. They wish to control the next election so that a sufficient number of Maggots get elected and everyone else can piss off. Does anyone really believe that if la Presidenta is still in power in 3.5 years that he'll give up power? He has no intention of running for a third term, he has every intention of making third term a fait accompli, no election necessary.
Musk needs government contracts (Score:2)
Tesla is collapsing and SpaceX is doomed if we ever have another Fair election. Musk interfered with a war and people are going to remember that. If the cronies he bought off with his government money ever get unseated the first thing they're going to do is spin up an alternative to SpaceX.
The rest of his businesses are either a stock market pumping dump or u
It's coming for your job... (Score:4, Insightful)
Editor/Copywriter. "Good enough" is perfect when it'll only get a fraction of second of brain time. LLMs are already laying waste to marketing departments around the world. It'll write "all" the product copy, just copypasta the user manual with some additional info, like pricing, sales promotions, etc.
Anyone still saying the gains are negligible, has their head firmly planted in the sand, imo. Jobs far below your pay-grade aren't about to be decimated (1/10), no, more hung and quartered (4/10). Anything involving marketing, sales, writing and storytelling in general, a lot less people will be able to make a living from this.
There's going to be a hell of a lot of people out of work soon. We're heading into a employers market, a long one. For the first time in my life, I'm feeling the "prepper" vibe. "Shit's about to get very real"...but they've been saying that forever, haven't they?
Re: (Score:3)
The demographic shift is removing far more employees in other jobs. I'm not saying the people can just change jobs, just that the changes within the context of the general clusterfuck of current day is not that impressive.
Until true AI, it's just one more thing.
We've been here before (Score:3)
It's capable (Score:2)
Just get it done (Score:3)
It takes significantly more time to describe accurately what you want than actually doing it. If you allow any level of ambiguity in the AI prompt, it is guaranteed that it AI will fuck something up. So you're spending more time refining the prompt and then fixing the mess, making it work, than you would without AI.