If you include microsoft copilot trials in fortune 500s, absolutely. A lot of major listed companies are still oblivious to the functionality of AI, their senior management don't even use it out of laziness
There is probably a threshold effect above which the technology begins to be very useful for production (other than faking school assignments, one-off-scripts, spam, language translation, and political propaganda), but I guess we're not there yet. I'm not counting out the possibility of researchers finding a way to add long term memory or stronger reasoning abilities, which would change the game in a very disorienting way, but that would likely mean a change of architecture or a very capable hybrid tool.
My company’s behind the curve, just got nudged today that I should make sure my AI use numbers aren’t low enough to stand out or I may have a bad time. Reckon we’re minimum six months from “oh whoops that was a waste of money”, maybe even a year. (Unless the AI market very publicly crashes first)
My compsci brain suggests large orgs are a distributed system running on faulty hardware (humans) with high network latency (communication). The individual people (CPUs) are plenty fast, we just waste time in meetings, or waiting for approval, or a lot of tasks can't be parallelized, etc. Before upgrading, you need to know if you're I/O Bound vs CPU Bound.
The thing with a lot of white collar work is that the thinking/talking is often the majority of the work… unlike coding, where thinking is (or, used to be, pre-agent) a smaller percentage of the time consumed. Writing the software, which is essentially working through how to implement the thought, used to take a much larger percentage of the overall time consumed from thought to completion.
Other white collar business/bullshit job (ala Graeber) work is meeting with people, “aligning expectations”, getting consensus, making slides/decks to communicate those thoughts, thinking about market positioning, etc.
Maybe tools like Cowork can help to find files, identify tickets, pull in information, write Excel formulas, etc.
What’s different about coding is no one actually cares about code as output from a business standpoint. The code is the end destination for decided business processes. I think, for that reason, that code is uniquely well adapted to LLM takeover.
But I’m not so sure about other white-collar jobs. If anything, AI tooling just makes everyone move faster. But an LLM automating a new feature release and drafting a press release and hopping on a sales call to sell the product is (IMO) further off than turning a detailed prompt into a fully functional codebase autonomously.
These surveys don’t make sense. Ask the forward thinking companies and they’ll say the opposite. The flood of anti AI productivity articles almost feel like they’re meant to lull the population into not seeing what’s about to happen to employment.
It’s funny because at work we have paid Codex and Claude but I rarely find a use for it, yet I pay for the $200 Max plan for personal stuff and will use it for hours!
So I’m not even in the “it’s useless” camp, but it’s frankly only situationally useful outside of new greenfield stuff. Maybe that is the problem?
Just to be clear, the article is NOT criticizing this. To the contrary, it's presenting it as expected, thanks to Solow's productivity paradox [1].
Which is that information technology similarly (and seemingly shockingly) didn't produce any net economic gains in the 1970's or 1980's despite all the computerization. It wasn't until the mid-to-late 1990's that information technology finally started to show clear benefit to the economy overall.
The reason is that investing in IT was very expensive, there were lots of wasted efforts, and it took a long time for the benefits to outweigh the costs across the entire economy.
And so we should expect AI to look the same -- it's helping lots of people, but it's also costing an extraordinary amount of money, and the few people it's helping is currently at least outweighed by the people wasting time with it and its expense. But, we should recognize that it's very early days, and that productivity will rise with time, and costs will come down, as we learn to integrate it with best practices.
Maybe! Or it might never pan out, or it may pan out way better.
Complicated things like this rarely turn out the way people expect, no matter how smart.
I mean the productivity paradox was only temporarily remedied. Around 2005 we entered a second version of the paradox and it persists to this day. I'll note that 2005 was when the internet became dominated by walled-gardens and social-media, _and_ it was the last year that people got to use the internet without smartphones (in 2006 LG released a smartphone, with Apple releasing iPhone in 2007).
The combination of attention-draining social media walled gardens, and the high performance pocket-computers (which are really designed for consumption instead of productivity), created a positive feedback loop that helped destroy the productivity that we won by defeating the paradox in the 1990s. And we have been struggling against this new paradox for twenty years, since. AI seems like it should defeat the paradox because it is a kind of hands-free system, perfect for mobile phones -- but this is really just a very expensive solution to a problem that we have created and allowed to fester. We could just shun the walled gardens, and demand to be paid for our attention and data.
The new productivity paradox (which I do not think AI in its current form can fix[1][2]), is the price that we pay for a prosperous and valuable advertising industry. And as long as the web is seen as an ad-channel, and as long as the web is always vibrating in your pocket, we will keep paying this price. We will eventually end up (metaphorically) lobotomizing our children, and families, and communities, so that the grand-children of ad-executives and tech-bros and frat-bros can grow up healthy, psychologically stable, educated, and comfortably wealthy. (Brain drain: now available literally everywhere).
[1]: It is telling that most LLMs are centralized, and are most useful as search-engines/information-retrieval-systems. The centralization makes them _spyware_, and their ability to directly answer any question, encourages users to actually ask direct questions, instead of stringing search-terms together. This makes the prompts high-signal advertising data (i.e. instead inferring what you are looking for from the search-string, these companies can see _exactly_ what you are looking for and why -- and with LLMs, they can probably turn these promps into joint-probability-tables or whatever other kind of serialization they need to figure out which products to sell you (either on the web or directly in the response to your prompt)).
[2]: As far as copyright infringement goes, LLM outputs may require mass clean-room rewrites (so your productivity, as pathetic as it already is, now gets _halved_ long term) of text, prose, code, and anything else that is produced with them, because of how copyright law works. In legal arts this is called _the fruit of the poison tree_, and any short-term productivity gains, may become long term liabilities that need to be replaced due to _legal mandate_ -- so even if LLMs can eventually produce perfect and faultless outputs, the copyright laws _in all 200+ countries_ would have to be torn down and rebuilt (and this will certainly come at great expense).
I like AI and use it daily, but this bubble can’t pop soon enough so we can all return to normally scheduled programming.
CEOs are now on the downside of the hype curve.
They went from “Get me some of that AI!” after first hearing about it, to “Why are we not seeing any savings? Shut this boondoggle down!” now that we’re a few years into bubble, the business math isn’t working, and they only see burning piles of cash.
Figure A6 on page 45: Current and expected AI adoption by industry
Figure A11 on page 51: Realised and expected impacts of AI on employment
by industry
Figure A12 on page 52: Realised and expected impacts of AI on productivity
by industry
These seem to roughly line up with my expectations that the more customer facing or physical product your industry is, the lower the usage and impact of AI. (construction, retail)
A little bit surprising is "Accom & Food" being 4th highest for productivity impact in A12. I wonder how they are using it.
It's not just technology, it's very hard to detect the effect of inventions in general on productivity. There was a paper pointing out that the invention of the steam engine was basically invisible in the productivity statistics:
The article suggests that AI-related productivity gains could follow a J-curve. An initial decline, as initially happened with IT, followed by an exponential surge. They admit this is heavily dependent on the real value AI provides.
However, there's another factor. The J-curve for IT happened in a different era. No matter when you jumped on the bandwagon, things just kept getting faster, easier, and cheaper. Moore's law was relentless. The exponential growth phase of the J-curve for AI, if there is one, is going to be heavily damped by the enshitification phase of the winning AI companies. They are currently incurring massive debt in order to gain an edge on their competition. Whatever companies are left standing in a couple of years are going to have to raise the funds to service and pay back that debt. The investment required to compete in AI is so massive that cheaper competition may not arise, and a small number of (or single) winner could put anyone dependent on AI into a financial bind. Will growth really be exponential if this happens and the benefits aren't clearly worth it?
The best possible outcome may be for the bubble to pop, the current batch of AI companies to go bankrupt, and for AI capability to be built back better and cheaper as computation becomes cheaper.
At $dayjob GenAI has been shoved into every workflow and it's a constant source of noise and irritation, slop galore. I'm so close to walking away from the industry to resume being a mechanic, what a complete shit show.
The people who will be most productive with AI will be the entreprompteurs who whip up entire products and go to market faster than ever before, iterating at dangerous speeds. Lean Startup methodology on pure steroids basically.
Unfortunately I think most of the stuff they make will be shit, but they will build it very productively.
It’s simple calculus for business leaders: admit they’re laying off workers because the fundamentals are bad and spook investors, admit they’re laying off workers because the economy is bad and anger the administration, or just say it’s AI making roles unnecessary and hope for the best.
It's weird being on here and seeing so much naysaying, because I see a radical change already happening in software development. The future is here, it's just not equally distributed.
In the past 6 months, I've gone from Copilot to Cursor to Conductor. It's really the shift to Conductor that convinced me that I crossed into a new reality of software work. It is now possible to code at a scale dramatically higher than before.
This has not yet translated into shipping at far higher magnitude. There are still big friction points and bottlenecks. Some will need to be resolved with technology, others will need organizational solutions.
But this is crystal clear to me: there is a clear path to companies getting software value to the end customer much more rapidly.
I would compare the ongoing revolution to the advent of the Web for software delivery. When features didn't have to be scheduled for release in physical shipments, it unlocked radically different approaches to product development, most clearly illustrated in The Agile Manifesto. You could also do real-time experiments to optimize product outcomes.
I'm not here to say that this is all going to be OK. It won't be for a lot of people. Some companies are going to make tremendous mistakes and generate tremendous waste. Many of the concerns around GenAI are deadly,serious.
But I also have zero doubt that the companies that most effectively embrace the new possibilities are going to run circles around their competition.
It's a weird feeling when people argue against me in this, because I've seen too much. It's like arguing with flat-earthers. I've never personally circumnavigated Antarctica, but me being wrong would invalidate so many facts my frame of reality depends on.
To me, the question isn't about the capabilities of the technology. It's whether we actually want the future it unlocks. That's the discussion I wish we were having. Even if it's hard for me to see what choice there is. Capitalism and geopolitical competition are incredible forces to reckon with, and AI is being driven hard by both.
As we approach the singularity things will be more noisy and things will make less and less sense as rapid change can look like chaos from inside the system. I recommend folks just take a deep breath, and just take a look around you. Regardless on your stance if the singularity is real, if AI will revolutionize everything or not, just forget all that noise. just look around you and ask yourself if things are seeming more or less chaotic, are you able to predict better or worse on what is going to happen? how far can your predictions land you now versus lets say 10 or 20 years ago? Conflicting signals is exactly how all of this looks. one account is saying its the end of the world another is saying nothing ever changes and everything is the same as it always was....
I think the 'AI productivity gap' is mostly a state management problem. Even with great models, you burn so much time just manually syncing context between different agents or chat sessions.
Until the handoff tax is lower than the cost of just doing it yourself, the ROI isn't going to be there for most engineering workflows.
I read an article in FT just a couple days ago claiming that increased productivity was becoming visible in economic data
> My own updated analysis suggests a US productivity increase of roughly 2.7 per cent for 2025. This is a near doubling from the sluggish 1.4 per cent annual average that characterised the past decade.
And, hear me out here - perhaps for the sake of morale it makes sense to leave a smidge of the part of the job that actually attracts people to this profession in the first place on their plates. Otherwise we may find that, after the novelty wears off, we’re left with a net productivity dropoff because there’s not as much left to keep people motivated to do à good job of the remaining work.
111 comments
[ 4.4 ms ] story [ 77.0 ms ] threadOther white collar business/bullshit job (ala Graeber) work is meeting with people, “aligning expectations”, getting consensus, making slides/decks to communicate those thoughts, thinking about market positioning, etc.
Maybe tools like Cowork can help to find files, identify tickets, pull in information, write Excel formulas, etc.
What’s different about coding is no one actually cares about code as output from a business standpoint. The code is the end destination for decided business processes. I think, for that reason, that code is uniquely well adapted to LLM takeover.
But I’m not so sure about other white-collar jobs. If anything, AI tooling just makes everyone move faster. But an LLM automating a new feature release and drafting a press release and hopping on a sales call to sell the product is (IMO) further off than turning a detailed prompt into a fully functional codebase autonomously.
So I’m not even in the “it’s useless” camp, but it’s frankly only situationally useful outside of new greenfield stuff. Maybe that is the problem?
Which is that information technology similarly (and seemingly shockingly) didn't produce any net economic gains in the 1970's or 1980's despite all the computerization. It wasn't until the mid-to-late 1990's that information technology finally started to show clear benefit to the economy overall.
The reason is that investing in IT was very expensive, there were lots of wasted efforts, and it took a long time for the benefits to outweigh the costs across the entire economy.
And so we should expect AI to look the same -- it's helping lots of people, but it's also costing an extraordinary amount of money, and the few people it's helping is currently at least outweighed by the people wasting time with it and its expense. But, we should recognize that it's very early days, and that productivity will rise with time, and costs will come down, as we learn to integrate it with best practices.
[1] https://en.wikipedia.org/wiki/Productivity_paradox
Wide spread internet access turned expensive toys (PCs) into useful assets.
Maybe! Or it might never pan out, or it may pan out way better. Complicated things like this rarely turn out the way people expect, no matter how smart.
The combination of attention-draining social media walled gardens, and the high performance pocket-computers (which are really designed for consumption instead of productivity), created a positive feedback loop that helped destroy the productivity that we won by defeating the paradox in the 1990s. And we have been struggling against this new paradox for twenty years, since. AI seems like it should defeat the paradox because it is a kind of hands-free system, perfect for mobile phones -- but this is really just a very expensive solution to a problem that we have created and allowed to fester. We could just shun the walled gardens, and demand to be paid for our attention and data.
The new productivity paradox (which I do not think AI in its current form can fix[1][2]), is the price that we pay for a prosperous and valuable advertising industry. And as long as the web is seen as an ad-channel, and as long as the web is always vibrating in your pocket, we will keep paying this price. We will eventually end up (metaphorically) lobotomizing our children, and families, and communities, so that the grand-children of ad-executives and tech-bros and frat-bros can grow up healthy, psychologically stable, educated, and comfortably wealthy. (Brain drain: now available literally everywhere).
[1]: It is telling that most LLMs are centralized, and are most useful as search-engines/information-retrieval-systems. The centralization makes them _spyware_, and their ability to directly answer any question, encourages users to actually ask direct questions, instead of stringing search-terms together. This makes the prompts high-signal advertising data (i.e. instead inferring what you are looking for from the search-string, these companies can see _exactly_ what you are looking for and why -- and with LLMs, they can probably turn these promps into joint-probability-tables or whatever other kind of serialization they need to figure out which products to sell you (either on the web or directly in the response to your prompt)).
[2]: As far as copyright infringement goes, LLM outputs may require mass clean-room rewrites (so your productivity, as pathetic as it already is, now gets _halved_ long term) of text, prose, code, and anything else that is produced with them, because of how copyright law works. In legal arts this is called _the fruit of the poison tree_, and any short-term productivity gains, may become long term liabilities that need to be replaced due to _legal mandate_ -- so even if LLMs can eventually produce perfect and faultless outputs, the copyright laws _in all 200+ countries_ would have to be torn down and rebuilt (and this will certainly come at great expense).
CEOs are now on the downside of the hype curve.
They went from “Get me some of that AI!” after first hearing about it, to “Why are we not seeing any savings? Shut this boondoggle down!” now that we’re a few years into bubble, the business math isn’t working, and they only see burning piles of cash.
Figure A6 on page 45: Current and expected AI adoption by industry
Figure A11 on page 51: Realised and expected impacts of AI on employment by industry
Figure A12 on page 52: Realised and expected impacts of AI on productivity by industry
These seem to roughly line up with my expectations that the more customer facing or physical product your industry is, the lower the usage and impact of AI. (construction, retail)
A little bit surprising is "Accom & Food" being 4th highest for productivity impact in A12. I wonder how they are using it.
https://www.frbsf.org/wp-content/uploads/crafts.pdf
However, there's another factor. The J-curve for IT happened in a different era. No matter when you jumped on the bandwagon, things just kept getting faster, easier, and cheaper. Moore's law was relentless. The exponential growth phase of the J-curve for AI, if there is one, is going to be heavily damped by the enshitification phase of the winning AI companies. They are currently incurring massive debt in order to gain an edge on their competition. Whatever companies are left standing in a couple of years are going to have to raise the funds to service and pay back that debt. The investment required to compete in AI is so massive that cheaper competition may not arise, and a small number of (or single) winner could put anyone dependent on AI into a financial bind. Will growth really be exponential if this happens and the benefits aren't clearly worth it?
The best possible outcome may be for the bubble to pop, the current batch of AI companies to go bankrupt, and for AI capability to be built back better and cheaper as computation becomes cheaper.
Unfortunately I think most of the stuff they make will be shit, but they will build it very productively.
In the past 6 months, I've gone from Copilot to Cursor to Conductor. It's really the shift to Conductor that convinced me that I crossed into a new reality of software work. It is now possible to code at a scale dramatically higher than before.
This has not yet translated into shipping at far higher magnitude. There are still big friction points and bottlenecks. Some will need to be resolved with technology, others will need organizational solutions.
But this is crystal clear to me: there is a clear path to companies getting software value to the end customer much more rapidly.
I would compare the ongoing revolution to the advent of the Web for software delivery. When features didn't have to be scheduled for release in physical shipments, it unlocked radically different approaches to product development, most clearly illustrated in The Agile Manifesto. You could also do real-time experiments to optimize product outcomes.
I'm not here to say that this is all going to be OK. It won't be for a lot of people. Some companies are going to make tremendous mistakes and generate tremendous waste. Many of the concerns around GenAI are deadly,serious.
But I also have zero doubt that the companies that most effectively embrace the new possibilities are going to run circles around their competition.
It's a weird feeling when people argue against me in this, because I've seen too much. It's like arguing with flat-earthers. I've never personally circumnavigated Antarctica, but me being wrong would invalidate so many facts my frame of reality depends on.
To me, the question isn't about the capabilities of the technology. It's whether we actually want the future it unlocks. That's the discussion I wish we were having. Even if it's hard for me to see what choice there is. Capitalism and geopolitical competition are incredible forces to reckon with, and AI is being driven hard by both.
Until the handoff tax is lower than the cost of just doing it yourself, the ROI isn't going to be there for most engineering workflows.
> My own updated analysis suggests a US productivity increase of roughly 2.7 per cent for 2025. This is a near doubling from the sluggish 1.4 per cent annual average that characterised the past decade.
good for 3 clicks: https://giftarticle.ft.com/giftarticle/actions/redeem/97861f...
- reviews for code
- asking stakeholders opinions
- SDLC latency (things taking forever to test)
- tickets
- documentations/diagrams
- presentations
Many of these require review. The review hell doesn't magically stop at Open source projects. These things happen internally too.