LLMs in their current state are very useful writing tools but not necessarily good thinking tools. A programmer can be more productive because boilerplate can be generated with a prompt like other 'templating' procedures but it's just far more flexible.
I recall a professor in my university stating that there is a vast difference between knowing information and having access to information. While this distinction in the era of the internet, search engines and smart phones seems slight the distinction is becoming more and more important.
It's worth stating because we are 'knowledge' workers and pride ourselves on our ability to learn and adapt. Every technological advancement has made certain jobs redundant and yet the prophecies that people will sit around idling and twidling their thumbs has never come close to fruition.
There will always be 'work' because there are things we want to do and things we don't want to really do and work is simply the latter. We all wish to work purely to buy our future laziness even 'security' can be understood in this sense.
It will be intersting to see how our learning institutions adapt -- pretty soon CS grads will have classes on how LLMs work in theory and more so in practice -- how to prompt effectively.
I'm less fearful than many -- this is just as someone put it before 'auotmation on steroids'.
> It's worth stating because we are 'knowledge' workers and pride ourselves on our ability to learn and adapt.
I don't think this is universally true for people working in IT. For some it definitely is, others don't care about their work and avoid learning and adapting at all cost because it's effort. They want the paycheck, but you'll need to threaten them with termination to get them to learn something new.
Those regular paycheck jobs exist but they’re never going to be safe and they’ve never been safe. I’ve been in tech for decades and waves of safe jobs disappearing happens regularly as technology improves.
I view these types of articles as 2025’s version of the 2003 article below, which was very big news in its day.
Outsourcing didn’t cause all American programming jobs to go away but it did raise the bar. Expect the bar for “professional programmer” to go higher still. And expect to level up your skills if you haven’t already.
On the flip side, the number of amateur programmers is likely to increase a lot with improved tooling.
Unfortunately one of the most important aspects of writing is that it helps you think clearly. A writing tool that sidesteps that part of the process will lead to deeply confused writings and people.
I find the 'rubber ducking' process of writing a hard problem down in to a concrete well articulated question is actually useful, even if the response is nonsense (which it is, maybe half the time.)
And who's to say CEOs actually know what they're doing? If anything, considering that most companies fail, relying on the opinions of a few cherry-picked CEOs doesn't hold much weight IMHO.
> While there is no direct evidence linking all these layoffs to AI, the trend is happening during a period of record economic strength.
The tech sector accounts for just about 6% of the U.S. workforce and roughly 10% of GDP. It's a stretch to draw broad conclusions about the economic impact of AI based on the comments of a few tech CEOs, especially when they represent such a narrow slice of the economy.
As usual, articles like this are more about sensationalism than substance.
Seriously, you always think of Germany as this technologically advanced country, but reality sets in pretty fast once you live here. With respect to IT, Germany is a few years behind the US, maybe as much a 8-10 years. They've also for years been looking (rather unsucessfully) to attract more academics to relocate here.
So, if you're willing to move continents, consider it. Just beware that the salaries are substantially lower for tech professionals - see parenthetical above.
Salaries are lower, taxes and cost of living are relatively high. Companies are much more reluctant to hire because of how difficult it is to let people go. For larger companies, you need credentials to have a good chance at being hired. Outside of (expensive) metropolitan areas, English isn't too easy to get by with (forget about handling bureaucratic tasks in English).
Can be good if you're fine with that, but it's definitely not as rosy as you make it out to be. And it's also not everyone's cup of tea to work in technologically underdeveloped economies and deal with fax machines.
> you always think of Germany as this technologically advanced country
Do you? Germany is notorious for being a bit behind in many sectors of tech, notably financial stuff. They’re still figuring out the whole cash vs card thing - while in surrounding countries people don’t carry wallets anymore.
That specific example is not (only) because of technological issues though, it's also the way Germans feel about banks and trusting them. In general, people are a lot more worried about their privacy here, too.
Not sure if anyone paying attention is surprised. US industry in particular has been anti labor openly since the 1980s after whether Reagan did during the FAA strike, its only gotten worse over time, even in tech.
There was a time where software companies at least did seem, if you read the historical records available, treat workers actually well on balance. Like at Google, there were secretaries that became millionaires because they were granted stock options (a practice that has gotten less generous over time and more restrictive)
I feel like from about 1996-2012 was the golden years, everything got corporatized as the MBAs flooded in around 2013.
I’m talking about early Google, Netscape, early Facebook, Sun Microsystems, flurry of Web 2.0 companies etc
>In 2024, 551 tech companies laid off nearly 152,922 employees, according to data from Layoff.fyi. The pace has accelerated dramatically this year. In just the first six months of 2025, 151 tech companies have already laid off over 63,823 people. On average a tech company cut 277 workers in 2024. If that rate is maintained for the rest of the year, the average number of layoffs per tech company in 2025 would soar to 851, roughly three times the 2024 average.
Am I missing something with this analysis? It seems like 2025 is on pace for fewer workers laid off, not more.
Yeah, but statistics don't work like that. Are those companies a representative sample? Did only 150 companies lay workers off in 2025? How did they extrapolate for 2025?
This isn't an area where you can trust the direct statements because everyone's looking for a way to cut costs without harming investor's faith in the potential for growth.
Yes, that's 100% correct. They'd be trumpeting increased productivity numbers from the rooftops if LLMs were actually changing things that much. It's recessionary cost-cutting masked with tech hype.
No disagreements with the article. Best to be clear eyed as this comes down the pipe. Getting told 'we will fire you as soon as we can' is not a great workplace, but the bigger indictment of management here is that they have no plan for growth. If your people are more productive, you could invest that in more ambitious projects if you had any good ideas about new opportunities.
You look at Andy Jassy's company wide memo referenced in the article and it starts with 'we are the world's largest start up', followed by claims of increased productivity that will be applied toward cost cutting instead of growth.
Most VC's would pass on a 'start up' that has such an apparent no/slow growth mentality.
The cold comfort for folks displaced by AI at major employers is that there’s still a glut of smaller ones that don’t use AI (either due to affordability, politics, or AI just can’t do what humans can). The issue with LLMs as the current AI-du-jour is that while they might replace the copy-pasters at Big Tech companies, these same companies are also assuming that AI is desired or usable in every other company or organization as well - which it’s not.
Once you deviate away from Big Tech stuff, their products just don’t integrate with the rest of the market very well. Manufacturing concerns might still be migrating to IPv4 and Ethernet PLCs, small businesses might not have the budget for AI Agent creation and retraining, and medium enterprises might be too bespoke to use AI effectively. That doesn't even touch the glut of companies who still do work on paper, not digitally, and who still need assistance with their digital transformation.
They’re calculating that their toys can eventually replace 80% of workers, and man they’re wrong.
I have this impression as well. AI, interest rates are secondary. Tech industry has just run out of ideas that would promise profits - and that didn't start in 2023.
Same story as with "overhiring". If you have a bigger crew, you just go plunder bigger ships - except if the ships are just not there or you have no idea how to find them.
> Tech industry has just run out of ideas that would promise profits - and that didn't start in 2023.
This, and it’s not getting enough attention.
Big Tech hasn’t had ideas other than acquisitions for decades. Startups aren’t inventing new things, just honing existing tech into an acquirable moat. Their big moonshots have failed one by one, with their two remaining ones - AI (current) and Quantum (next) - on their last legs of patience with shareholders. NFTs, Crypto, the “Metaverse”, Big Data, the Cloud, all of these have had varying degrees of initial success but now see scrutiny in the face of dwindling funds and rising costs, which is forcing Big Tech to try and adapt to being mature entities rather than startups.
It extends from B2B into the B2C realm as well. Yearly refreshes are increasingly underwhelming as components stop scaling like they did in the past, bumping up against physical limitations of materials that can’t be overcome without substantial R&D. R&D that they don’t want to pay for, because it’s high risk and high reward (a scenario tech isn’t really used to anymore). Nobody’s swinging for the fences, just incrementing very slowly over time until someone else does the heavy lifting who they can poach in an acquisition.
It's not just that, but it's also that more and more people are getting into the industry for its promise of high paying relatively cushy jobs. Just look at university CS major enrollments. Not to mention bootcamps and accelerated programs.
I've been feeling like this has been coming long before AI doomsayers were a thing.
The problem is that it's surprisingly difficult to judge candidates from the interview process, so the expectation value of a new hire has a lot of uncertainty. And while degrees from good schools are a good indicator, tech jobs perhaps unlike medical or legal fields, for example, are not the end all be all for success. It's entirely possible to be smart and self motivated and taught.
Interest rates matter, and having a global reserve currency helps capital markets that fund these things.
I would suggest 3 other forces:
- saturation: we don’t have any legacy fields left that aren’t yet digitized. If you I go from encyclopedia -> search engine or taxi -> uber you have a first mover green field market
- dominance games: VCs want scale, and scale is difficult with niche markets. Yet there are innovations there, but who cares? A new shitcoin has the potential to scale quickly, whereas startups in even energy seems to be a downer for them
- no manufacturing: long iteration cycles on anything that needs hardware. Plus, lower margins, and the Chinese will sell cheap knockoffs anyway if you’re successful, so bigger risk. Thus, more shitcoin and AI wrappers.
It's not that there are no ideas just no one willing to invest in anything that takes longer than 1 year or involves hardware. Most all new unicorn businesses are just rent seeking platforms. There's absolutely zero interest in creating value, only capturing it from someone else.
From what I’ve heard cream of the crop grads in hot fields are still getting huge offers and getting snapped up by the big names as well as smaller firms.
Some of those must be real positions doing promising work.
I don't understand why GenAI hasn't replaced middle management yet. One VP should be able to use GenAI stuff to manage hundreds+ of direct reports. They'd be able to ask agents for status updates whenever that part of their gray matter wakes up, and there'd be less meetings for everyone.
I’ve seen this take and it just feels like it’s coming from an IC who resents their manager or doesn’t understand the role.
Thing is there are plenty of useless managers. There are also useless ICs and useless C-suite execs. GenAI might replace them all.
A good middle manager (any manager really) needs to have a degree of empathy to keep people motivated. They need to read beyond the status updates to understand what’s actually happening.
You can easily replace junior engineers with genAI because most of that role was automated training the models from all the stuff junior engineers read.
It’s going to be a disaster if you think posting status updates to a glorified slack bot is going to give you the same outcomes as a good manager who knows how to manage humans.
I can't my finger on exactly why, but I have the strong feeling that having a liable human is greatly desired. Perhaps because upper management themselves don't want to be liable for managing an AI.
A lot of the AI purging is a dressing up the ridiculous over hiring at these big companies and anticipating the economic malaise to come. This fall is going to be a bloodbath.
These plans are not about cutting moribund departments that produce nothing. These are plans to fire people who are currently performing well and delivering for the company absent any downturn. This is very different than cutting bloat in that these people are currently holding up their end of the bargain for employment. Personally I would also suggest that seeking gains that come 100% at the expense of a business partner is a real asshole move.
HN also used to ridicule organizations that have no vision for future growth. From a business perspective the more scathing indictment of these executives is that they have no plan for capitalizing on AI. If a VC asked a startup what they would do with an extra X engineers or $Y in cloud credits, and the answer is "we would do the same things we are doing now" the VC would probably pass on the company. AI gives you twice the engineers. If your company doesn't have any plans to take advantage of this opportunity then it is already in decline.
Old skool HN was about starting companies. Most of the startups are pretty meh these days and everyone wants to go hide in some massive company that is printing money.
The big growth era is over. We’re entering an era of conglomerates and bullshit government back scratching. The big tech companies are going to be like 1960’s ITT, FMC, etc.
I have to reluctantly say that Claude Code has done a lot of work in more than a week that I could in double time. But doing so it made a lot of mistakes.
I'm not really sure about the final opinion I should make. It's fast for sure but it's far from correct. Sometimes it just feels a placebo.
It's a mixed bag to the point where I would agree that it's placebo. It is great for getting you from nothing to something quickly but it makes no many mistakes that I think it evens out to being marginally faster than if I did it on my own at the cost of my own learning and knowledge to where I don't use it any more.
I think it’s decent for getting a project bootstrapped (which is always a slog), but the further you get beyond that and the more you’re writing app-specific logic, the less useful it becomes.
That’s valuable, particularly for those who have trouble with pushing projects through those earliest phases, but it’s hardly a panacea.
Readit News
I recall a professor in my university stating that there is a vast difference between knowing information and having access to information. While this distinction in the era of the internet, search engines and smart phones seems slight the distinction is becoming more and more important.
It's worth stating because we are 'knowledge' workers and pride ourselves on our ability to learn and adapt. Every technological advancement has made certain jobs redundant and yet the prophecies that people will sit around idling and twidling their thumbs has never come close to fruition.
There will always be 'work' because there are things we want to do and things we don't want to really do and work is simply the latter. We all wish to work purely to buy our future laziness even 'security' can be understood in this sense.
It will be intersting to see how our learning institutions adapt -- pretty soon CS grads will have classes on how LLMs work in theory and more so in practice -- how to prompt effectively.
I'm less fearful than many -- this is just as someone put it before 'auotmation on steroids'.
I don't think this is universally true for people working in IT. For some it definitely is, others don't care about their work and avoid learning and adapting at all cost because it's effort. They want the paycheck, but you'll need to threaten them with termination to get them to learn something new.
I view these types of articles as 2025’s version of the 2003 article below, which was very big news in its day.
Outsourcing didn’t cause all American programming jobs to go away but it did raise the bar. Expect the bar for “professional programmer” to go higher still. And expect to level up your skills if you haven’t already.
On the flip side, the number of amateur programmers is likely to increase a lot with improved tooling.
https://www.cbsnews.com/news/training-your-own-replacement/
> While there is no direct evidence linking all these layoffs to AI, the trend is happening during a period of record economic strength.
The tech sector accounts for just about 6% of the U.S. workforce and roughly 10% of GDP. It's a stretch to draw broad conclusions about the economic impact of AI based on the comments of a few tech CEOs, especially when they represent such a narrow slice of the economy.
As usual, articles like this are more about sensationalism than substance.
Seriously, you always think of Germany as this technologically advanced country, but reality sets in pretty fast once you live here. With respect to IT, Germany is a few years behind the US, maybe as much a 8-10 years. They've also for years been looking (rather unsucessfully) to attract more academics to relocate here.
So, if you're willing to move continents, consider it. Just beware that the salaries are substantially lower for tech professionals - see parenthetical above.
Can be good if you're fine with that, but it's definitely not as rosy as you make it out to be. And it's also not everyone's cup of tea to work in technologically underdeveloped economies and deal with fax machines.
Do you? Germany is notorious for being a bit behind in many sectors of tech, notably financial stuff. They’re still figuring out the whole cash vs card thing - while in surrounding countries people don’t carry wallets anymore.
Especially in tech. Tech has always been anti-labor and anti-union.
I feel like from about 1996-2012 was the golden years, everything got corporatized as the MBAs flooded in around 2013.
I’m talking about early Google, Netscape, early Facebook, Sun Microsystems, flurry of Web 2.0 companies etc
Am I missing something with this analysis? It seems like 2025 is on pace for fewer workers laid off, not more.
You look at Andy Jassy's company wide memo referenced in the article and it starts with 'we are the world's largest start up', followed by claims of increased productivity that will be applied toward cost cutting instead of growth. Most VC's would pass on a 'start up' that has such an apparent no/slow growth mentality.
Once you deviate away from Big Tech stuff, their products just don’t integrate with the rest of the market very well. Manufacturing concerns might still be migrating to IPv4 and Ethernet PLCs, small businesses might not have the budget for AI Agent creation and retraining, and medium enterprises might be too bespoke to use AI effectively. That doesn't even touch the glut of companies who still do work on paper, not digitally, and who still need assistance with their digital transformation.
They’re calculating that their toys can eventually replace 80% of workers, and man they’re wrong.
Are there really 'copy-pasters' at 'BIg Tech' companies?
Same story as with "overhiring". If you have a bigger crew, you just go plunder bigger ships - except if the ships are just not there or you have no idea how to find them.
This, and it’s not getting enough attention.
Big Tech hasn’t had ideas other than acquisitions for decades. Startups aren’t inventing new things, just honing existing tech into an acquirable moat. Their big moonshots have failed one by one, with their two remaining ones - AI (current) and Quantum (next) - on their last legs of patience with shareholders. NFTs, Crypto, the “Metaverse”, Big Data, the Cloud, all of these have had varying degrees of initial success but now see scrutiny in the face of dwindling funds and rising costs, which is forcing Big Tech to try and adapt to being mature entities rather than startups.
It extends from B2B into the B2C realm as well. Yearly refreshes are increasingly underwhelming as components stop scaling like they did in the past, bumping up against physical limitations of materials that can’t be overcome without substantial R&D. R&D that they don’t want to pay for, because it’s high risk and high reward (a scenario tech isn’t really used to anymore). Nobody’s swinging for the fences, just incrementing very slowly over time until someone else does the heavy lifting who they can poach in an acquisition.
It’s bad, ya’ll.
I've been feeling like this has been coming long before AI doomsayers were a thing.
The problem is that it's surprisingly difficult to judge candidates from the interview process, so the expectation value of a new hire has a lot of uncertainty. And while degrees from good schools are a good indicator, tech jobs perhaps unlike medical or legal fields, for example, are not the end all be all for success. It's entirely possible to be smart and self motivated and taught.
I would suggest 3 other forces:
- saturation: we don’t have any legacy fields left that aren’t yet digitized. If you I go from encyclopedia -> search engine or taxi -> uber you have a first mover green field market
- dominance games: VCs want scale, and scale is difficult with niche markets. Yet there are innovations there, but who cares? A new shitcoin has the potential to scale quickly, whereas startups in even energy seems to be a downer for them
- no manufacturing: long iteration cycles on anything that needs hardware. Plus, lower margins, and the Chinese will sell cheap knockoffs anyway if you’re successful, so bigger risk. Thus, more shitcoin and AI wrappers.
From what I’ve heard cream of the crop grads in hot fields are still getting huge offers and getting snapped up by the big names as well as smaller firms.
Some of those must be real positions doing promising work.
Thing is there are plenty of useless managers. There are also useless ICs and useless C-suite execs. GenAI might replace them all.
A good middle manager (any manager really) needs to have a degree of empathy to keep people motivated. They need to read beyond the status updates to understand what’s actually happening.
You can easily replace junior engineers with genAI because most of that role was automated training the models from all the stuff junior engineers read.
It’s going to be a disaster if you think posting status updates to a glorified slack bot is going to give you the same outcomes as a good manager who knows how to manage humans.
Ignore all the talk about 'the flattening' . middle management is back with a vengeance, bigger than ever, at meta and airbnb.
A lot of the AI purging is a dressing up the ridiculous over hiring at these big companies and anticipating the economic malaise to come. This fall is going to be a bloodbath.
Note how often product managers or engineering managers appear in layoffs.
Just happened to a neighbor.
Deleted Comment
HN used to ridicule bloated organizations. “Why is there 1000 people? That’s a weekend project plus support.”
Now it defends them.
These plans are not about cutting moribund departments that produce nothing. These are plans to fire people who are currently performing well and delivering for the company absent any downturn. This is very different than cutting bloat in that these people are currently holding up their end of the bargain for employment. Personally I would also suggest that seeking gains that come 100% at the expense of a business partner is a real asshole move.
HN also used to ridicule organizations that have no vision for future growth. From a business perspective the more scathing indictment of these executives is that they have no plan for capitalizing on AI. If a VC asked a startup what they would do with an extra X engineers or $Y in cloud credits, and the answer is "we would do the same things we are doing now" the VC would probably pass on the company. AI gives you twice the engineers. If your company doesn't have any plans to take advantage of this opportunity then it is already in decline.
The big growth era is over. We’re entering an era of conglomerates and bullshit government back scratching. The big tech companies are going to be like 1960’s ITT, FMC, etc.
That’s valuable, particularly for those who have trouble with pushing projects through those earliest phases, but it’s hardly a panacea.