everybody believes every sector bounces back until a specific one doesn’t and stays gone for 20+ years.
even then, someone can get elected and cake up by promising to get it back.
i was in the exhausted, abandoned Appalachian coal fields in the 2010s and there were plenty of small business tyrants in ghost towns eager to believe coal was just about to come back when obama ruined everything.
its not to say many, if not most, totally knew shit was over and never coming back. but the true believers tended to be in charge of shit, threw their weight around more, and acted like naysayers were gonna undermine the comeback with their lack of faith.
Same shit in Maine. There was a woman up there who was buying a bunch of land in one area in order to donate it to the National Park system, and it pissed off a bunch of people who thought she was going to kill the paper mill comeback that they insisted was just around the corner.
They really missed that paper mill smell
As devastating as the pulp mill shutdown a few decades ago was to my old hometown, the air has never been cleaner. The mill was within walking distance of the downtown core because it started over a century ago and workers had to live within walking distance. Private automobiles were the height of luxury back then.
im honestly even skeptical the AI bubble popping will make AI go away. I think we’re more likely to see a 2000s situation where the bottom falls out, the actually dominant companies remain, and we find ourselves dealing with the one or two companies appointed to carry it forward. Consolidation, not outright destruction.
which depresses the unholy fuck out of me
the way power and capital are all in on it conceptually - it just does too many things that serve them - they want the skilled labor elimination, they want the information control, they want the thing to mature enough to be an entertainment replacement.
i frankly half expect, a decade from now, seeing AI companies buying out entertainment companies and industries and just shuttering them so that people have to use their product. e.x.: buying up instrument creators so you have to either be wealthy or capitulate and use a LLM to do that labor for you.
the future is bleak, and the way they want that technology to develop, and the sheer amount of money and “were forcing this down your throats” that’s happening in several places and in several sectors of the economy - i absolutely despise i have to talk to an AI chatbot to apply to certain jobs now, but that’s psychologically conditioning the workforce to accept and use it at the point of a gun, and it’ll work over time if it is allowed to continue.
I am pretty sure tech is about to devour itself entirely, leaving everything even more consolidated and what’s left even more entrenched in our lives, with more power to influence and demand presence in the zeitgeist, no matter what the average person wants.
Like, yall, we need to be thinking really hard about what that technology is intended to do and the logical consequences of it persisting. Do not sit and hope the old normal will return. Historically this has never fucking happened. The .com crash did not kill the internet, it just heralded the end of it as a frontier.
You should ask those people in what sense is the market going to come back up and what the timeline is.
Is it going to go back up in the sense that the oversupply of CS graduates will be solved? Probably.
Is it going to go back up in the sense that CS graduates will be given 6 figure salaries and VIP treatment? 0% chance.
How will the oversupply be solved?
The salaries will become low enough that people will give up on getting a Cs degree
I agree with this. A big chunk of the supply was not from fresh 4-year graduates. There are several postbacc programs in the US that target career switchers, which are already seeing their enrollments tank within the last year or so. People are less willing to spend more money on a 2nd degree if they already have a stable but undesirable career if it no longer seems that software engineers get significantly higher pay.
If nothing else, eventually people will leave the field, and fewer new grads will enter the field. Might take 20 years though, who knows.
As a current nerd, I don’t know where our future nerds will go instead. I went from maths to a software factory full of mech/chem/electronic engineering grads who realized at around graduation time that there’s not a whole lot of engineering happening in the US. With that outlet shrinking, will
s start proliferating?
It may be cope.
I do think tech is more likely to “bounce back” than a lot of other industries. Like some people here are comparing it to extractive industries that never came back but I think demand for tech innovation is probably going to remain a lot more consistent (baring societal collapse) than demand for coal or timber or whatever.
I do think it will continue to be a less prestigious and well compensated field, as more people learn tech skills employers will have less need to lavish talent with high pay and job perks.
I fundamentally believe we can’t engineer software anymore. Like how we can’t manufacture anything anymore. So I don’t think there will be more tech growth in western nations. A good number of my high tech friends work for companies based in other nations. Open source software for my field is mostly Chinese contribution. I’m in the late stages of the interview process for an Indian company. I live in Silicon Valley.
I work in tech and I think this trend is one of the funniest things. The better / more complete your software is the less likely it is to succeed. The software that does the best is the one that is the shiniest. It doesn’t have to work. Working is actually a detriment to it because time spent making it work is less time spent making it shiny.
The shit I have seen be successful these days is obscene. Entire companies are making bank on an AI tool that gets it wrong half the time and doesn’t even know it. The people that have to deal with it just…don’t. And none of it seems to matter. It’s data moving around so much for so little reason at all that if some of the data is wrong or fake it just doesn’t matter. What I think will really pop the bubble is when suddenly it does matter; the mountain of fake or wrong data that entire industries are being built on will completely collapse. And it doesn’t have to be that way; people could actually make software and make it work but there’s no incentive to do that because the time you spent making software good is time someone else spent selling a promise that will never be.
The companies that make good products still will survive the crash if they can survive this phase. Doubly so if they can figure out how to fix the damage. Not the first time tech has gone that way and probably won’t be the last.
For the last decade or so, the tech jobs market has been dominated by big tech hiring everyone they can, so they can’t wander off on their own and make new competitors. That’s predicated on big tech having so much money that they can’t find anything more useful to spend it all on. It won’t continue past the tech growth stock bubble, whenever it pops. (Which might not be the pop we’re currently headed for; if it’s just AI instead of the whole thing somehow, then those jobs will bounce back.)
On the positive side, if that ever stops, some of the unemployed tech workers will make actually useful things, some of which will become businesses that can hire the other tech workers. This is more like how it worked in the early 2000s.
However, those new tech companies have to actually make money somehow, which means somebody has to be able to afford their products, which means new tech will be beholden to the shape of the wider economy. For example, currently startups are all B2B, because businesses have money and consumers don’t. In a wide scale economic downturn where businesses are cutting costs, even those B2B startups won’t be able to find customers, so they’ll have to fold (or just not get founded).
So I don’t think the market is going to recover to the old maximum, and afterwards it’ll be much more cyclical, following the rest of the economy along whatever nonsense it does.
I imagine that the amount of useful activities which can be performed on a computer remaining which haven’t been implemented in some form of application or digital service is virtually zero. The tech industry boomed in the 90s because consumer access to PC hardware and the Internet was skyrocketing. It boomed again in the 00s-10s because smartphones filled a niche which computers couldn’t, and consumer access skyrocketed further. Today, the market is saturated and there are no meaningful hardware innovations left to make, no new apps to write.
Of course, all this existing software and IT infrastructure requires maintenance, but the era of “upstart” VC-backed apps/websites becoming multi-billion dollar firms is finished. With “AI,” we’ve reached the mountaintop removal phase of a dying industry.
In hindsight, the writing was already on the wall when they started introducing shit like smart ovens and washing machines. They already ran out of runway.
Maybe all of the stuff that needs to be written has been written, but the defacto standards are often not the best choices and they are almost always behind a huge corporate paywall. So if someone can figure out how to make money off of open source (don’t laugh!) there’s a huge niche there
Personally I think it’s a bit of cope, but it’s also realistic that there is a bounce back.
Unless there is a societal collapse, there will always be a tech industry. However, not to the level it was. I think it will be more of a correction than anything.
Depends on a few things.
- Does the definition of “tech sector” broaden to essentially the entire economy? Hard to think of a large company that doesn’t have some kind of tech department outside of IT at this point. However, when people say “tech sector” they’re mostly talking about the big FAANG (or whatever the acronym is now) companies where you make lots of money working for a somewhat prestigious brand. Safe bet is the number of jobs requiring tech skills will increase overall, but probably won’t pay as well.
- Does AI actually pan out well enough to replace coders within the next few years? If so it will eat entry-level programming jobs first, but I think it will be quite a while before AI is effectively writing OS or database code further down the stack. Thus the task for tech workers becomes specializing faster than AI can. However, the issue with specialized jobs is that there are fewer of them. And they are more difficult, technically.
- People are bad at predicting markets and are almost always cursed to become “bagholders” (here, of a CS degree presumably) perpetually waiting for the market to hit again. Psychologically it is very difficult to break out of this state and accept the good times aren’t coming back, due to the FOMO of bailing right before the long-awaited gigantic green candle. However, the analogy breaks down here because while skills do decay with lack of practice, the fundamental programming skill is more like riding a bike - you don’t forget. Thus if you retrain as an electrician or something it isn’t unthinkable you could switch back to programming when times get better.
Absolutely nobody could have predicted the current state of things based on Attention Is All You Need. Tech moves so quickly and unpredictably that trying to predict another 7 years is an easy way to make a fool of yourself. Someone could introduce an alternative to transformer architecture that makes AGI possible, Google Glass might catch on for real this time, companies might start generating content to increase engagement, as opposed to just recommending it, photonic chips might create a breakthrough in compute, or some secret other thing that nobody realized. Personally, I see robotics as a field with a lot of achievable progress to be made. Self driving cars already exist, and the market for them is real. Airlines are looking at single pilot cockpits, the demand and theoretical feasibility for fully autonomous forklifts exist, and tons of other things. The world hasn’t run out of software to build yet.
You now basically have a basket-weaving degree.
LLMs are good enough at coding now to replace the average fresh comp sci graduate. You still need a couple senior people to check the output, but tech companies don’t need nearly as many grunts as they used to. It’s very possible that AI will plateau where it is now for a long time and the people with experience get to keep their jobs, but it’s gonna be brutal for newcomers from now on.
LLMs are good enough at coding now to replace the average fresh comp sci graduate.
This really isn’t true. Modern LLMs are still not much better than an advanced Google search. I’m not even in CS doing industry work, but even I can spot misses in LLM output.
On a basic level, the AI has 2 major disadvantages. Firstly, it is not fully upto date with the Internet (and training on data past 2022 risks poisoning the dataset), a problem that will get worse over time. Secondly, LLMs are expensive as hell to actually run. Thirdly, the LLM context windows and higher order reasoning are still limited.
Firstly, it is not fully up to date with the Internet (and training on data past 2022 risks poisoning the dataset).
Where on earth did you get that from? Sonnet-4.5 has a pre-training cutoff date of January 2025 and GPT-5 has a pre-training cutoff date of October 2024. Any vaguely modern interface can get data past that into context by RAG and MCP. These aren’t far back because of model collapse or anything, it’s just that fine tuning is a hugely labor intensive process that takes months. Model collapse is greatly mitigated with human-based feedback and finetuning, making it safe to train models on LLM generated data. Deepseek, for example, is directly trained off GPT and Claude’s output.
I am aware that LLMs do train with datasets past 2022. But there is a risk of poisoning the dataset that will grow overt time as the use of LLMs becomes larger. It is not a risk that can be easily mitigated by human feedback and fine tuning, since getting rid of workers is exactly why business owners are hyped about LLMs in the first place.
And yes, I did not about MCP so I was wrong about that part, but you can still put less data into context vs in training.
LLMs are good enough at coding now to replace the average fresh comp sci graduate
This depends heavily on the graduate. I’ve met fresh graduates who don’t know what a for loop is and I’ve met fresh graduates who have a pretty extensive portfolio already. It really depends on how much they engaged with their program, hung out with other people from their major, and did projects outside of schoolwork.
All that said, junior developers are capable of doing something that LLMs can’t currently do: becoming senior developers. If the industry wants to do this short sighted bullshit of giving juniors the shaft after all the work they put into “learn 2 code” they’re going to be painfully surprised when their seniors start leaving and there’s no labor pool to replace them.
If the industry wants to do this short sighted bullshit of giving juniors the shaft after all the work they put into “learn 2 code” they’re going to be painfully surprised when their seniors start leaving and there’s no labor pool to replace them.
That’s exactly what I would expect to happen, they never plan more than a quarter or two in advance
This really isn’t true for any non-trivial project. If you want a basic CRUD app maybe, maybe. But anything more complicated or novel than that and the LLM isn’t doing much aside from boilerplate for you.
Even for a CRUD app, it’s questionable. Ironically, we’ve been able to scaffold those pretty easily since the early aughts without burning a small rain forest in the process.
