AI Can Technically Perform 12% of US Labor Market's Wage Value, MIT Simulation Finds (cnbc.com)
- Reference: 0180222129
- News link: https://news.slashdot.org/story/25/11/27/0752210/ai-can-technically-perform-12-of-us-labor-markets-wage-value-mit-simulation-finds
- Source link: https://www.cnbc.com/2025/11/26/mit-study-finds-ai-can-already-replace-11point7percent-of-us-workforce.html
The researchers call this the Iceberg Index, and the name is deliberate. The visible AI disruption happening in tech jobs right now accounts for only 2.2% of labor market wage value. The remaining exposure lurks in cognitive and administrative work across finance, healthcare administration, and professional services, and unlike tech-sector disruption, it's spread across all fifty states rather than concentrated on the coasts.
Delaware and South Dakota show higher Iceberg Index values than California because their economies lean heavily on administrative and financial work. Ohio and Tennessee register modest tech-sector exposure but substantial hidden risk in the white-collar functions that support their manufacturing bases.
To validate the framework, the researchers compared their predictions against Anthropic's Economic Index tracking real-world AI usage from millions of Claude users. The two measures agreed on state categorizations 69% of the time, with particularly strong alignment at the extremes.
The Iceberg Index doesn't predict job losses or adoption timelines. It measures technical capability, the overlap between what AI can do and what occupations require. Traditional economic indicators like GDP and unemployment explain less than five percent of the variation in this skill-based exposure, which is partly why the researchers argue workforce planners need new metrics.
[1] https://iceberg.mit.edu/report.pdf
20% (Score:4, Informative)
20%: The amount of their time actual people will have to spend to fix the mistakes made by AIs doing 12% of the work.
Re: (Score:2)
There's a few restaurants in town that have those little robot waiters. And they *all* have a human following them around.
I save time asking an LLM how to code X (Score:4, Informative)
But I still have to try the solutions out and get into my code. All in all, I save time because I don't have to look up solutions around the web myself. But I always have to code it myself, except in one case whete I got it to translate existing (obscure) code to Python - but not error free either, it turned out.
Re: I save time asking an LLM how to code X (Score:3)
I've recently been using Chatgpt for assistance with typescript and css for a work SPA project. One time I asked it if a very small function - only 5 or 6 lines - was the most efficient, idiomatic way to do a thing. It gave me a response, then asked if I wanted an even "more efficient" way - which, duh, of course I do, that's what I asked. The "more efficient" way was a literal copy of my original function.
I despise CSS so much that I'm sure I will continue to use LLMs to tell me why a layout isn't working,
Analog (Score:2)
AI is only able to do 12% of digitized work. Recently I bought a tablet with a stylus. I'm doing a course so I made notes. There seems to be no AI that can pick out even one word from my notes. I think we are forgetting that humans occasionally need to interact with things that humans have not already encoded into 1s and 0s. Furthermore, AI is only good if it can work with digitized information.
My recent experience vibe coding (Score:4, Interesting)
I have some medium-sized open source projects that I write or contribute to on github.
1) my daughter's form-based web project--loads of content, loads of pages with various inputs. She had only partial content, and ChatGPT took it upon itself to fill in the rest. The content was quite good. I never did get satisfactory layout, however.
2) refactoring of a god-object from another repo. It took awhile, but Claude got there much faster than I would have. Added additional functionality at my request. Now published and working. Generated documentation of classes and methods.
3) conversion of a jsx website to a tax (JavaScript to typescript). I figured this would be a disaster, but nope. Claude did the conversion (about 50 components and additional methods) in a few days working with it. Also, all documentation, including release notes. And tests.
My experience:
1) treat the agent like a talented junior developer who is very fast and quite thorough.
2) it will get confused and forget things. An instruction.md file really helps to prevent regressions
3) it will get stuck in loops and go down rabbit holes at times. Test and commit often so you can rolllback breakage.
4) proceed incrementally where possible. Small, discrete steps work best
5) ask the agent to analyse/explain before doing
6) don't be afraid to ask it for suggestions; they can be quite good (it did a nice job improving the layout and color scheme of my website, for instance).
AI is like a retarded child (Score:1)
AI has some uses, but I feel it's still at the retarded child phase. Like any child, I would trust it to function on its own without supervision. Since it's also retarded, it actually requires extra supervision, since when AI freaks out it will really shit the bed and create a horrible mess that humans will need to spend triple the time (or more) to cleanup....
DId you know? (Score:1)
If we'd just kill everybody, 100% of the labor market's value could be replaced by no people at all. Isn't that amazing? It also is by all legitimate logic and rules oif extrapolation and deduction the ultimate goal of our economical and political system. I don'tz want to say I am a fucking prophet or something but I think THE END IS NIGH!
Infinite ego engine (Score:2)
As long the AI thinks it is always right and never self check or correct, it's not truly "ready" for anything.
Specially when the failure mode is "as close as possible to the right answer" because it works by "the statistically more probable answer".
It's like a digital vegeta, but even worse.
Yeah, maybe, in theory (Score:4, Informative)
In practice, it's more likely that the best workers will use AI to become more productive, while carefully reviewing its work and correcting if needed
Yes, this will reduce the need for some poorer quality workers, but it's silly to believe that today's AI can operate unsupervised
So 12%... (Score:2)
... of people now working and earning a living (and, to be real, it's the US) and having medical coverage can all be put out of work, evicted, to die under bridges, while the CEOs increase their ROI.
What have I missed?
It will be everywhere (Score:2)
It's about to replace all porn actresses as well.
The AI bubble (Score:3, Interesting)
This is why everyone and their grandmother is all in on AI. It's adoption lags for the sole reason of "people haven't caught up with what it can do, and learned how to let it do it".
Whatever investment is currently in AI is dwarfed by what value it can already technically do, as long as people actually integrate it. It's the "get people to integrate it" part that is the choke point.
Meanwhile AI and robotics advancing rapidly is constantly increasing the amount of work it can perform adequately (about as well or better than typical human worker in that job).
As has been said before, primary problem with AI is no longer model quality, or robotics quality. It's human adaptation to "these tasks can be done by AI as well or better than a human, so we should develop workflow where these tasks are handed off to AI so humans can focus on other things that AI can't do".
Meanwhile models and robotics continue to rapidly advance, so this remains a moving target, making it even harder to hit at any given point in time.
Notably, by far the biggest all in on AI right now is in PRC, where entire factories are going to AI, most of research work is already AI assisted (hence the demand for 4090s and 5090s that have enough VRAM for narrow models commonly used in research), and things like robotics are among top if not the top of the world. Everyone else is behind.
Re: The AI bubble (Score:2, Informative)
None of the "models" can reliably and repeatably do what they are claimed to do. Humans are still the only reliable choice, and there is absolutely no evidence that this will not be the case for the foreseeable future.
Re: The AI bubble (Score:4, Insightful)
the hunger by the 1% to remove as much humanity from the workplace is sickening.
they fully know they are destroying the middle and lower classes (even more than they already have).
they, like the R party, just dont care. they think they are rich and insulated enough. they never cared what their own people need. the 'let them eat cake' time has come back again, but even worse.
there will be no thought to social systems needed to support the unemployed (which will be many of us, given enough time).
I'm glad I'm retiring soon. I would not want to compete in a job market that bosses think can be done by computer, alone.
and I would not want to be the 'prompt meister' to try to coax answers from the machines that make sense.
some see a great future with AI. I see nothing but doom and gloom. the greed factor is strong in humans and the class disparity will cause rioting and civil wars.
maybe not wars. the US has created a special police force that is above the law, so any uprisings will EASILY be dealt with. they thought about that. ICE is not just for foreigners. its a general purpose police force answerable only to 1 person.
people, please show me I'm wrong. but all signs point to a very bad future for 95% of the 'thinks for a living' workforce.
Re: (Score:2)
> the hunger by the 1% to remove as much humanity from the workplace is sickening.
To be fair, this is nothing new. The entirety of the industrial revolution has been finding ways to use automation to decrease the amount of human labor used to make things (i.e., increase "productivity".)
The problem is that we do not have an economic system in which a society works when there is no need for human labor, and a small but rich fraction of the population owns the machinery that produces everything.
Re: (Score:1)
You can choose to reject much of the industrial revolution. Most Westerners are able to purchase human-crafted personal goods. From 100% re-built autos to hand-woven suits and dresses, the items are available. The price? Consumption of a fewer number of "long term" purchases, and great self-satisfaction in identifying master-craft products. As a rule, let peons and sociopaths buy mass-produced items.
Can craft save the economic system [Re: The AI...] (Score:2)
>> The entirety of the industrial revolution has been finding ways to use automation to decrease the amount of human labor used to make things (i.e., increase "productivity".) The problem is that we do not have an economic system in which a society works when there is no need for human labor, and a small but rich fraction of the population owns the machinery that produces everything.
> You can choose to reject much of the industrial revolution. Most Westerners are able to purchase human-crafted personal goods. From 100% re-built autos to hand-woven suits and dresses, the items are available. The price? Consumption of a fewer number of "long term" purchases, and great self-satisfaction in identifying master-craft products.
You can choose a lot of different things. The question remains, is this a viable way to structure an economic system in a world in which all of the necessities of life are produced with no (or almost no) labor?
Are you seriously proposing a world in which eight billion people are employed in producing master-crafted articles (and these master-crafted articles are "long term" purchases, hence with a small output needed.)?
> As a rule, let peons and sociopaths buy mass-produced items.
Where do the peons get the money to buy mass-produced items?
A handful are master craftsme
Re: (Score:1)
We know what will happen in a world where there is no need for human labour. The 'elite' will build Terminators to eliminate most of the humans.
They are not going to pay for billions of people to sit around watching porn all day.
Re: (Score:2)
We do. It's got to do with socuakising the means of production and the main reason it hasn't been done lately is that it will require some literal wading in the blood of the oppressors as they'll be better protected than Louis XIV and his pals. As soon as the masses really have nothing left to lose, however, that won't help,
Re: (Score:2)
To me, AI is like the discovery of the electric motor. It can affect everything across our entire life. However, the key is how it is used. When the electric motor was invented, people didn't tout that it would make people turning wrenches obsolete. It was made as a very useful tool for stuff at the time, and was a huge improvement in factories from the old "big steam motor in the basement, with tons of gears and belts...".
We also have had AI for decades. Genetic algorithms, fuzzy logic, etc. It is on