Today my employer had their quarterly state of the company conference call. For a while now, the CEO has been pushing the use of AI, and on the call they made the statement “Anyone who isn’t using ChatGPT should start now, or they won’t be part of this company in the future.”
Has anyone else experienced an actual threat to their jobs based on AI? Over use of it, or being actually replaced by it?
I work for a major cable provider doing repair over the phone. If your internet or cable TV goes out, I’m the one you talk to. AI is already integrated in that our calls are now transcribed and we no longer keep notes, as it actually summarizes our calls. Our tool that guides troubleshooting used to be kind of a guide, but lately we’ve been forced to follow it exactly. It’s fairly obvious that it is being trained via our experience.
Yes. The job fundamentally no longer exists, because AI could do it at a perceived fraction of the cost (though they didn’t pay enough) with lower but adequate accuracy. And since it was part of a federal program, my prior employer exited the market figuring between AI being able to do it, and funding wars being a now annual/bi-annual thing, that it wasn’t worth it.
In my industry (web dev), it’s a big deal and a double-edged sword. Programming is one of those things AI is pretty good at, and hallucinations aren’t as big a deal because you can instantly test the output for correctness. I, and other humans in the field, have already seen a lot of displacement from AI, especially for junior coders. Entry level roles are extremely rare now, especially compared to the glut during Covid.
More than that, a lot of traditional software jobs and companies are also not receiving investment anymore, with all the money and mindshare going to AI players instead.
Now traditional software seems like a dead-end career, and we’re the new coal miners. If and when I lose my current job, I don’t think I’ll be able to find another one with the same skills again. I’ve been doing this for 30+ years, and I’ve never seen it be this bad. AI will kill this profession.
Despite that, myself and most other devs I know are all-in on AI, even as it threatens our jobs. It’s just too good at it, and in many ways is a much better coder than I ever was or could be. It just has access to decades of coding education and examples that no human could memorize. The only things it still can’t do very well right now (in terms of this profession) are product management and user experience. I still have to babysit it a lot, but it’s already substantially better than most interns, juniors, Google, or traditional software help sites. And it’s only getting better every month.
Overnight my job has changed from coder to mentor, where I rarely directly write code anymore, but instead guide along an AI with English prompts while waiting for it to code, and then verifying and correcting its output with more prompts. It’s an incredible force multiplier when supervised by an experienced human developer. But we all know it’s only a matter of time before we’re replaced altogether.
But that’s OK. Jobs come and go. I’ve been better off, I’ve been worse off. Such is the nature of technology under capitalism. And honestly, web dev in the US was severely overpaid for being such a low to mid skill profession, and I’m glad to see these course corrections, even if they endanger my own livelihood. I’m also really intrigued by the potential for longer term good that could come from this technology, even as it disrupts labor and markets in the shorter term. It’s going to be a terrifying few years and decades, but maybe a few generations from now, it’ll just be an indispensable part of modern life, the way smartphones are today.
I’m retired now, but there has been a long term push to drive in this direction in my biz too. There’s not much doubt it’ll start appearing for real around the edges of commercial aviation here within 10 years. Experiments and prototypes are flying today.
The problem I always foresaw was that it would be comparatively easy to slightly retrain a skilled experienced journeyman worker to supervise an AI “apprentice” rather than a human apprentice. Such as you describe yourself doing.
But when that guy retires, where does the next generation of human journeyman come from?
I’m not sure if the apt metaphor is eating the seed corn or inching out on a limb then sawing it off behind you. But in either case it amounts to designing a system to consume the valuable paid-for resource that is accumulated human journeyman skill with no thought given to how to replace that resource once it’s used up.
The blithe assumption that the journeymen too will become obsolete and will be replaced by even better AIs at just the rate required is just that: blithe.
As an aside, my son is a recent grad (well, 1.5 years ago, but still recent), and he reports his cohort was encouraged to go into tech jobs. They graduated high school and entered college right when the pandemic caused a surge in tech jobs, so counselors were telling everyone to aim for a technical field. The washout of tech jobs as the pandemic wound down, at the same time these kids finished school, has been a disaster for them. My son chose and non-technical field (Finance) and has been doing good so far, but some of his friends are still working part-time/temp as the job market has had the bottom fall out, especially for new grads, and especially in technical fields. The rosy picture of abundant tech jobs appears to have been a mirage, and I am not sure it will ever get back to what it once was.
What would you tell a new grad with a degree in a technical field to do now, with the advent of AI?
Yeah, the pipeline is pretty fucked. With no juniors coming in anymore and the AIs not really being “trainable” in the traditional sense (you can’t truly teach them long-term skills unless you’re one of the huge companies with massive data centers doing the expensive training/fine-tuning), it’s a matter of time before they exhaust the supply of experienced devs.
But maybe that’s OK. In the same way that web devs of my generation didn’t necessarily need a comp-sci background to do our work, maybe web devs of the future won’t need to be coders at all. Today’s highly-abstracted dev environment is a pretty far cry from, say, having to learn assembly language, algorithms, or even HTML and CSS.
The overwhelming majority of my fellow web devs did not come from a comp sci program, and often did not come from formal secondary education at all, just some boot camp or online videos. It really often is a low-to-medium skill level profession, with a bunch of “paid amateurs” making use of a few frameworks/libraries made by the much smarter FAANG people.
Even today, already anybody can make an AI-assisted website using things like Wix AI, Squarespace AI, or the closer-to-the-code Vercel v0, and then refine it further with the LLM agent of their choice. Figma, the popular design suite, is also moving towards that direction. Eventually we’ll just get web agencies or lone wolves who can build entire small business sites with just these tools, similar to how it was in the 2000s, easily combining design, marketing, and the implementation just by describing things and drawing layouts.
Learning how to use AI effectively (as in how LLMs and transformers work, how contexts and prompts work, how tool calling etc. work) is already a pretty different skillset from traditional programming, anyway. And learning to actually train or fine-tune a transformer for your own needs is yet another skillset. It’s probably better for future generations to be trained directly on that layer instead of the underlying HTML/CSS/JS, all of which can and will be abstracted away.
Web dev is just one tiny microcosm. If it goes away as a profession, oh well, life goes on. But I feel like it’s just the canary in the coalmine.
It’s one thing to say oh, it’d be great for small businesses to be able to easily make their own websites again without needing to hire expensive developers (and even as a developer, I agree with that), but AI is threatening so many sectors of the economy all at once, not just web dev.
Which businesses would even survive the cataclysm, and who will be able to remain their customers? Rampant layoffs, stratospheric wealth inequality, the loss of basically all democratic checks and balances and consumer protections, runaway inflation, etc. are all joining forces into a perfect storm with AI and increasing automation.
I’m very scared for Gen Z and Alpha, and their kids especially. Today’s AI makes my job a little easier, but I think tomorrow’s AIs will completely reshape their career pipelines…
But, I dunno. The generations before mine probably felt the same way about the interweb and those computer things.
Overall I agree w your insights. Since the start of the Industrial Revolution, the future has always been a little scary to young folks due to changes coming down the pike.
But IMO the rate and quantity of change due to rampant AI deployment will inevitably create a major qualitative difference in the coping ability of individual people and the society they collectively embody.
The fact the politics of the next ~25-50 years in the so-called civilized world will also be uniquely uncivilized certainly increases the likelihood of a terrible outcome for almost everyone.
As an interpreter I can only add that I am glad my professional career is coming to a close. I enjoyed to good times and am sorry for the younger generations who will never be able to enjoy the privileges I had. The ones in their 40s and 50s today will be hardest hit: too old to learn another trade after investing a lot in what they do today but too young to retire.
My 2¢:
If they’re really smart and good with maths and statistics, they can directly tackle the primary research that’s ongoing (and exploding) in AI/ML right now, and become part of that tiny percentage of humanity actually able to help steer AI (and the rest of us) into the future. (That stuff is way, way above my head, personally, but YouTube has some nice gentle introductions and a longer course for anyone possibly curious.)
If they’re just ordinary smart, eh… maybe learn carpentry and Chinese?
Not even kidding… I think AI is going to massively accelerate the concentration of wealth towards the top and decimate the knowledge economy (not just devs, but writers, managers, analysts, creatives of all sorts), which the US is largely based on. Combined with our top-down (soon to be top-only) system of government, to me that means either you become part of the crème de la crème (in the Von Braun or Einstein sense) and thus shielded from the rat race, or you’re well and truly fucked, especially in the US where we we lack even basic social protections. Today’s job market is already ruthless, and I’d be surprised if it doesn’t become a literal bloodbath in the coming decades…
To that end, anything hands-on is probably safer than anything knowledge-based, at least in the short term (robotics keep getting better too, and it’s probably only a matter of time there as well). We’ll always need electricians and plumbers (if only for the data centers). My partner is a veterinary nurse and she’ll probably continue to be employed long after I’ve become homeless and destitute (hopefully she’ll let me crash with her…).
And although China will face many of the same problems we have today, their governance structure, relative homogeneity (and lack of/heavily suppressed racial violence), and heavy investment in both renewables and STEM will leave them better prepared for the next few decades, IMHO, compared to our nation’s current trajectory. What we still have for now is the bleeding edge of the research, still, but if our policies continue to become more and more xenophobic and nationalistic, European and Asian researchers will be less likely & able to come here too, perhaps leading to a reverse brain drain.
I don’t have much hope for our future here, regardless of field… if I were young and mobile, my priority would be trying to emigrate elsewhere ASAP before the rush. The problems we’re facing are uber-scale societal and generational issues, not questions of individual willpower or skillsets.
But that is strictly IMHO, and I’m sure there are people who are much more optimistic about our future.
A lot of project manager and middle manager types like myself are currently out of work, but I don’t think it’s directly due to AI.
A lot of consulting forms have also been shedding jobs. Accenture and Mckinsey have been in the news due to job cuts and Accenture is saying that the 11,000 people they laid off are those who are “untrainable” in AI. But that’s like 1.5% of their 700,000 employee workforce. Mckinsey was worse with about 10% of their 45,000 employees.
What I think is actually happening is that the overall economy sucks because of Trump’s mismanagement, tariffs, tech layoffs, and Federal layoffs. I think companies don’t have any idea how to increase revenue so everyone is glomming onto the AI bandwagon. Either hiding behind AI to hide downsizing due to normal cost reductions or trying to build up their perceived expertise.
The thing is, I’ve been studying AI recently (N8N, ChatGPT, etc) and before I was laid off our firm was pushing use of our Copilot accounts and AI as a “go to market” strategy. While some jobs like coding or customer service have been hit pretty hard, I just don’t see AI as this grand tool for making people 10x more productive. At least not yet. It helps with a lot of stuff, particularly tasks that are repetitive or workflow-oriented.
But what I do hate about AI is how it now makes it so easy for anyone to mass produce garbage, whether it’s crappy resumes, spammy outreach emails, marketing copy, creepy uncanny valley headshots, and whathave you. And most of that crap is by people trying to get you to buy in on AI related services.
Well not me, I build weird stuff for museums. AI can’t carpenter/ laminate/ install electronics/ hell today I was pounding stones into mortar for a new stream feature at our place, AI can’t do that…yet, until it builds hands. Or figure out on the fly how kids are going to break stuff (they’re amazing at breaking stuff.) That one I know AI will never get.
I guess doing physical construction of things, custom things, not mass production, will be safe for a while.
I don’t think the goal is to make people more productive, but instead to make companies more profitable by having AI do the work people used to do at a fraction of the cost. What all those unemployed people are supposed to do to earn money is not AI’s problem to solve (nor, apparently, those companies’ problem, either). If things get bad enough, then too many people wont have money to spend on the services AI is providing, then what?
Well, it’s not really the company’s problem to solve. The company’s job is to be profitable.
If we get to a point where a significant amount of work is automated away such that we can’t really find enough for people to do, that would require a significant restructuring of society. And I don’t just mean UBI. Even the very concept of a “city” as a place where we need to consolidate people to work on stuff would change.
The short answer is “I don’t know”. But I imagine the world my kids become adults in will look as different to me as our society looks to my grandparents.
This is my take as well. There’s hiring stagnation because of the weak economy and a even larger hesitation to invest because of uncertainty about the economy. When there actually is investment, it is AI related.
My daughter worked as a freelance copywriter, writing copy for web pages.
Once AI came along, they stopped using her.
Re AI replacing/supporting tech support for cable and internet problems:
I recently had an interaction with AI through chat for my internet service provider. To be fair, it was after-hours and I suspect that live staffing was either entirely absent or greatly reduced. I was quite surprised to find that the AI could not understand that I have been having intermittent problems with my service. Every few days, my internet connection (including the wifi) would drop for a period of up to two hours. A complete reboot of the router would sometimes fix the problem.
And I could not get the AI to help me, either by scheduling a tech to check the fiberoptic connection or to offer to replace the modem/router. It would keep running the routine script to check modem status, have me answer questions about the various indicator LEDs, and then reset the modem/router. Any comments I made pointing out that this was happening on a regular basis were completely ignored. Again, it never offered anything further…no offer of having a tech come out or even a follow-up call from a live support person. I was pretty shocked at its limitations.
That’s something well short of an LLM. Its an overgrown script reader w a couple branches in its script.
The more companies put these useless but cheap gatekeepers in place, the more troubleshooting and customer service in general becomes impossible for the customer.
I wonder how this ends? Badly I’m sure.
To be fair, that’s not that different from what a human working in first-level tech support for a cable company would say, lol.
At least the AI was trained on realistic data. “Stall and misdirect” is pretty much the MO of those companies.
Nobody has threatened my job with AI but my CEO keeps pushing me to use it to organize grants. I disagree with it on principle so I haven’t used it yet. The only thing I would really find it useful for would be automating my prospect research somehow, but I asked about it in a related thread and was told it can’t even help me with that.
I’m confident I’m a better writer than AI.
Eventually most grants will be written by AI but I’ll still be a better writer. It probably won’t affect my job because non-profits are touchy-feely and slow to change, but it might eventually make jobs like mine obsolete.
So far AI can’t write fiction for shit so I’m feeling pretty secure there. The only real issue is people have no taste anymore so they might not realize it’s shit. But I know.