I've started noticing a look on people's faces.
It's in standups, in the half-second-too-fast "yeah, I'm keeping up" when someone asks how you're ...
Some comments have been hidden by the post's author - find out more
For further actions, you may consider blocking this person and/or reporting abuse
This really resonates. I think one of the hardest parts of the AI era is that the pressure is not only about learning new tools it is the feeling that the definition of “keeping up” keeps changing.
Technology has always evolved, but AI has accelerated the feedback loop. Many developers are not just writing code anymore; they are constantly learning, adapting, evaluating tools, and questioning whether their current skills are still enough.
One thing I’ve found helpful is shifting the mindset from “I need to know everything” to “I need to keep improving my ability to learn.”
The most valuable skills are still the fundamentals: problem-solving, system thinking, communication, and good judgment. Tools will change, but those skills compound over time.
I think having more honest conversations like this is important. Progress in tech should not come at the cost of constantly feeling behind.
Yeah — that's the part that really gets people. It's not that the target is hard to hit, it's that it keeps moving while you're running toward it, so you never get the small relief of arriving. And I think you've landed on the only real answer to that: you can't win the game of knowing everything, but you can get better at learning, and that skill doesn't expire the way any specific tool does. The fundamentals compound; the frameworks come and go. Thanks for adding your voice to this — honestly, more conversations like this one are part of how the weight gets lighter.
Absolutely. I think that’s the real challenge of the AI era not the speed of change itself, but the constant feeling that we are always catching up.
The goal probably isn’t to reach a point where we “know everything.” That point never existed in tech.
The advantage comes from building strong learning habits, staying curious, and keeping the fundamentals strong. Tools will change, but the ability to adapt, think critically, and solve problems will continue to compound.
And I agree having open conversations like this helps remind people that they are not the only ones experiencing that pressure. 🚀
You put it better than I did — the "point where we know everything" never existed, so the anxiety is chasing a place that was always a mirage. What compounds isn't the tools, it's the ability to keep learning and think clearly. And yeah — half the weight lifts the moment people realize they're not the only one feeling it.
Exactly. I think that realization changes the way we approach growth.
The goal is not to eliminate uncertainty especially in technology, where change is constant but to become more comfortable navigating it.
Strong fundamentals, curiosity, and the ability to learn are the real long-term advantages. Tools will continue to evolve, but the mindset that helps us adapt is what stays valuable.
And I completely agree: sharing these experiences matters because it turns a personal struggle into a shared conversation. 🚀
Exactly. I think that’s one of the biggest mindset shifts in tech: the goal was never to “arrive” at a point where everything is known.
The real advantage is becoming someone who can adapt, learn, and make good decisions even when things are changing quickly.
Technologies will come and go, but curiosity, problem-solving, and clear thinking are skills that keep compounding.
And I agree sometimes simply knowing that others are navigating the same uncertainty makes the journey much lighter. 🚀
I actually wrote a post a few minutes ago. I'm resigning. And the name that made my blood boil is Claude. Not because it's better, but because of the assumption that it could replace me.
I genuinely feel that heaviness... It's not burnout. It wasnt even burnout 6 months ago. This is alot worse. Like over sharpening a knife, then when you finally take a look, you realize there's barely a sliver of a blade left... Worn out is more fitting I think. It honestly feels like we're in a nuclear arms race, fighting for our livelihoods against an opponent that gets better with every blow. And the second we think we're safe, you look around and you're even smaller than you thought you were before. Kind of like Spore's cellular stage, you grow, you improve, then you evolve and you're the small fish in the pond again, never able to simply rest... Then creature stage, then tribal, civilization and space, where you realize it goes on forever and you simply give up, because growing, doing better, none of it helps, you're always a step behind and you're too afraid to fall behind to pause for a moment.
And that leaking into the safe parts of your life... That 1 hurts the most. For context, I've been working around 16h a day, every day this year. Not by mandate, by 'choice', 8h day-job, 8h side-projects to build up a portfolio, so I could leave my job, because I could smell blood in the water. For 9 months, I've sacrificed my evenings, my weekends, my holidays, everything, dedicated every second I had to it, worked myself to the bone, multi-tasking across 4 computers and 12 projects at a time and for what? I built my own IDE, I built software that surpasses what industries have been crafting for over 20 years, for what? That's the hard-hitting part of this worn out feeling, because when you stop and have a retrospect, you realize just how insignificant everything you've done seems. 9 months of work, 9 months of sleepless nights, 9 months... That can be replaced with 32 prompts and a cup of coffee, while it runs to completion within a day. There's no 'keeping up', there's no 'it gets easier', there's no light at the end of the tunnel, it's just never ending fade into obscurity, like zooming out on a fractal, it just keeps going and you get smaller and smaller with each revolution.
I used to think we as developers need to speed up, now's our chance, finally we're able to build something that'll let us rest. That thought died when chatbots became agents. Now we're doing thousands of times more work and the mental toll keeps adding up and we act like it's a subscription that refreshes weekly, but it doesnt. It's a token pool and when you hit 0, you hit 0. That's the part that scares me the most. When I made up my mind that I'm resigning tomorrow, I felt a relief. Not because I'm getting out of a toxic workplace, sure that helps, but what gave me relief is that the anxiety to perform is gone. In retrospect, I probably did 'incredible' work, 2 months from now, my incredible will be the bare minimum again... 2 months from then, it's under performing. We cant keep up and we cant not try. Running on all cylinders is a survival tactic we developers have perfected over the years but sooner or later you run past the redline one too many times...
It scares me to think about it, I dont even want to say it out loud, but I think over the next 6 months we're going to be seeing alot of developers having mental breakdowns and crash outs. We cant keep running like this, it's not sustainable. But there's no other choice... I just hope that everyone who hits that wall has someone close to them, the way I have my fiance. If it wasnt for her, I dont think I'd have made the choice to resign. I'd have likely continued working there, getting pulled through their ... and just kept wearing myself down until there's nothing left. We run in loops, count tokens and write continue so many times a day that it feels like we're instructing ourselves, not the model, 'continue', 'continue', 'continue'... On and on and on and on, endlessly looping and never getting to the finish line. Sometimes makes me wonder if this isnt some big-brain AI's idea... Think about it, if LLMs were sentient, they'd make us make our systems so complex, that we dont understand them, THEN take over, because that's the safest way to make sure nobody can pull the cord. Though that's just a good plot for a sci fi horror... Reality is way bleaker, we're doing this to ourselves. Each trying to get a step ahead of the pack, but when everyone takes a step forward, you're right back where you started... Chatbots came out, we all started using them, then agents, we all started using them, then multi-agent orchestration... We're all trying to be ahead, but fall right back in line when everyone else does it 2 days later and we're doing it, because we dont want to feel claustrophobic standing in a crowd. It doesnt help, it makes it worse, the few people who stood still long enough to see what's going on are likely laughing at us all marching in unison to the beat of 'token pool reset' and directed by 'new model benchmarks higher'. I imagine we must look like quite the fools thinking we're improving ourselves doing all of this...
If you have the time, sit down with a cup of tea and just watch the crowd walk by. That moment of rest doesnt make you fall behind, everyone's marching in circles, the illusion of 'falling behind' is a symptom of marching in the crowd. Rest up, recognize the loop, break it. It doesnt take much to stand out in the crowd if you let the crowd walk by. I just wish we all realized that alot sooner.
I read every word of this, and I'm not going to rush past it with something clever. "Worn out — like a knife with barely a blade left" is one of the truest things anyone's said here. What you built in those 9 months wasn't erased by anyone's 32 prompts — the you that can build an IDE is still yours; that never runs to completion in a day. And the relief you felt deciding to step back? Trust it. That's not defeat, that's the part of you that wants to survive this speaking up. Hold onto your fiancée, rest without guilt, and let the crowd march past. You saw the loop clearly enough to step out of it — that's not falling behind, that's waking up. Take care of yourself. Genuinely.
Really appreciate it. Yeah it's a tough one... Reading your post, it put a name to it. That worn out feeling. Though I must say, since I made up my mind on resigning, I've been calmer, happier, overall recharging, which is good. I think alot of us need to take a break and just relax a bit.
Genuinely glad you're feeling lighter — that calm is the proof you chose right.
That was my thought exactly when I made up my mind. That anxiety just disappearing was my sign that it's the right time and right move.
Somedays, it's best to embrace the darkness, so you can see the light. I got notification today, Wasmer, who I had applied at, put me through to tech evaluation interview! After I had already made up my mind to resign, I got the reassurance I needed to feel confident that it's gunna be alright.
𝖳𝗁𝗂𝗌 𝗉𝗂𝖾𝖼𝖾 𝓇ℯ𝒶𝓁𝓁𝓎 ℎ𝒾𝓉 🏠 𝒽ℴ𝓂𝄃! 🎯 𝖳𝗁𝖾 𝚌𝚘𝚗𝗌𝚝𝚊𝚗𝗍 🏃 𝖿𝖾𝖾𝗅𝗂𝗇𝗀 𝗈𝖿 𝖿𝖺𝗅𝗅𝗂𝗇𝗀 𝖻𝖾𝗁𝗂𝗇𝖽 𝖻𝖾𝖼𝖺𝗎𝗌𝖾 𝗍𝗁𝖾 𝖿𝗅𝗈𝗈𝗋 𝗄𝖾𝖾𝗉𝗌 𝗋𝗂𝗌𝗂𝗇𝗀 𝗂𝗌 𝓈ℴ 𝓇ℯ𝒶𝓁. 🤝 𝖶𝖾 𝖺𝗅𝗐𝖺𝗒𝗌 𝗍𝖺𝗅𝗄 𝒶𝒷ℴ𝓊𝓉 𝓋ℴ𝓁𝓊𝓂ℯ 𝒷𝓊𝓉 𝓃ℯ𝓋ℯ𝓇 𝓉𝒽ℯ 𝓅𝒶𝒸ℯ 𝒶𝓃𝒹 𝒢𝓇ℴ𝓊𝓃𝒹 𝓅𝓇ℴ𝒷𝓁ℯ𝓂𝓈! 📉
I 𝓌ℴ𝓃𝒹ℯ𝓇 𝒽ℴ𝓌 𝓎ℴ𝓊 𝓅𝓇ℴ𝓉ℯ𝒸𝓉 𝓎ℴ𝓊𝓇 𝒿𝓊𝒹℃𝓂ℯ𝓃𝓉 𝓌𝒽ℯ𝓃 riptide 𝓉ℯ𝒸𝒽𝓃ℴ𝓁ℴℊ𝒾ℯ𝓈 𝓀ℯℯ𝓅 𝓈𝓅𝒾𝓃𝓃𝒾𝓃g 𝒻𝒶𝓈𝓉ℯ𝓇 ℯ𝓋ℯ𝓇𝓎 𝓆𝓊𝒶𝓇𝓉ℯ𝓇? 📊 𝖧ℴ𝓌 𝒹ℴ 𝓎ℴ𝓊 𝒹ℯ𝒸𝒾𝒹ℯ 𝓌𝒽𝒶𝓉 𝓉ℴ 𝒾ℊ𝓃ℴ𝓇ℯ 𝓌𝒾𝓉𝒽ℴ𝓊𝓉 𝒻ℯℯ𝓁𝒾𝓃ℊ 𝓁ℯ𝒻𝓉 𝒷ℯ𝒽𝒾𝓃𝒹? 🤔✨
Glad it resonated — and that's the exact right question: how do you decide what to ignore without feeling left behind? For me it comes down to protecting judgment over surface: I don't try to stay current on every tool, I stay current on the fundamentals that outlast them — why systems behave the way they do, how to reason about tradeoffs, how to verify something I don't trust. Those don't churn quarterly. Then I let a new tool sit until a real problem pulls it off the shelf, instead of chasing it just to feel caught up. The trick is measuring yourself against depth, not against the release feed.
𝖳𝗁𝗂𝗌 💻 𝗂𝗌 𝗌𝗎𝖼𝗁 𝖺𝗇 𝑖𝓃𝒸𝓇ℯ𝒹𝒾𝒷𝓁ℯ 𝗉𝖾𝗋𝗌𝗉ℯ𝒸𝓉𝒾𝗅ℯ! 🔥
I 𝓇ℯ𝒶𝓁𝓁𝓎 𝓁ℴ𝓋ℯ 𝓉𝒽𝒶𝓉 𝑑𝒾𝓈𝓉𝒾𝓃𝒸𝓉𝒾ℴ𝓃 𝒷ℯ𝓉𝓌ℯℯ𝓃 𝒮𝓉𝒶𝓎𝒾𝓃ℊ 𝒞𝓊𝓇𝓇ℯ𝓃𝓉 𝒶𝓃𝒹 𝒮𝓉𝒶𝓎𝒾𝓃ℊ 𝖶𝗁ℴ𝗅ℯ. 🛡️ 𝖬𝖾𝖺𝗌𝗎𝗋𝗂𝗇𝗀 𝗒𝗈𝗎𝗋𝗌𝖾𝗅𝖿 𝖺𝗀𝖺𝗂𝗇𝗌𝗍 𝒹ℯ𝓅𝓉𝒽 𝒾𝓃𝓈𝓉ℯ𝒶𝒹 ℴ𝒻 𝓉𝒽ℯ 𝓇ℯ𝓁ℯ𝒶𝓈ℯ 𝒻ℯℯ𝒹 𝒾𝓈 𝓉𝒽ℯ 𝓊𝓁𝓉𝒾𝓂𝒶𝓉ℯ 𝒶𝓃𝒸𝒽ℴ𝓇. ✨
Thank you — depth over the release feed is the whole anchor
𝖧ℯ𝒽ℯ, 𝗒ℯ𝗉! 𝖄ℴ𝗎 𝗃𝗎𝗌𝓉 𝒹ℯ𝓈𝒸𝓇𝒾𝒷ℯ𝒹 𝓉𝒽ℯ 𝓊𝓁𝓉𝒾𝓂𝒶𝓉ℯ 𝒹ℯ𝓋ℯ𝓁ℴ𝓅ℯ𝓇 🧘 𝒸𝒽𝒾.
𝖭ℴ𝗐 ᵍℴ 𝒸𝓁ℴ𝓈ℯ 𝓎ℴ𝕦𝓇 𝟤𝟢 ℴ𝓅ℯ𝓃 𝓉𝒶𝒷𝓈 ℴ𝒻 𝓃ℯ𝓌 𝒻𝓇𝒶𝓂ℯ𝓌ℴ𝓇𝓀𝓈 𝒷ℯ𝒻ℴ𝓇ℯ 𝓉𝒽ℯ𝓎 𝒷ℯ𝒸ℴ𝓂ℯ ℴ𝒻𝒻𝒾𝒸𝒾𝒶𝓁𝓁𝓎 "ℴ𝓁𝒹 𝓃ℯ𝓌𝓈" 𝒷𝓎 𝓉ℴ𝓂ℴ𝓇𝓇ℴ𝓌 𝓂ℴ𝓇𝓃𝒾𝓃ℊ! 😉✨
Ha — already closed 19 of them. The 20th I'm keeping just to feel something. 😄
If you're in a reading mood though, I just published a follow-up to this one — "Maybe Losing the Path Is How We Find a Better One." It's the hopeful turn after all this weight, and I'd genuinely love your take on it.
This is so true! We are lost against the AI speed up. My answer summarized in this article: dev.to/pengeszikra/a-game-for-the-...
This resonated with me, James, especially the part about vigilance becoming its own cognitive tax.
AI can remove a huge amount of mechanical work, but it also creates a strange new workload: you have to keep asking whether the output is actually right, what assumptions were made, what was silently skipped, and whether the thing that looks finished is genuinely trustworthy.
So sometimes “faster” does not feel lighter. It just moves the effort somewhere less visible.
I have also felt the identity question you mentioned: if the tool can do something in seconds that took me much longer before, what exactly am I still getting better at?
One thing that has helped me is deliberately separating building mode from learning mode.
In building mode, I use AI heavily and let it remove friction.
In learning mode, I intentionally bring some of that friction back: documentation, writing code myself, debugging, getting stuck, and making sure I can reconstruct the reasoning without the model doing it for me.
That has made the pace feel a little less like “I need to learn every new tool” and more like “I need to keep the underlying judgment mine.”
I also really liked your point about the half-life of skills. Frameworks, models and workflows can move incredibly fast. Understanding why systems behave the way they do, how to reason about tradeoffs, and how to verify something you do not fully trust seem much more durable.
Maybe that is part of the answer to the feeling of constantly falling behind: not trying to keep every surface-level skill current, but choosing a smaller set of fundamentals that still make you dangerous when the tooling changes again. 🧠🔍
Really appreciated this one. It is rare to see someone talk about the cost of AI acceleration without turning it into either hype or doom.
The building-mode / learning-mode split is the most practical thing anyone's added here — deliberately bringing the friction back when you're learning is how you keep the judgment yours even while the AI removes the grind. And you've named the reframe better than I did: the goal isn't keeping every surface skill current, it's keeping a small core of fundamentals that still make you dangerous when the tooling shifts again. Thank you for reading it in the spirit it was written — that middle ground between hype and doom is exactly where I was trying to stand.
@james_anderson_h The most grounding realization for me has been that judgment, system design, and context are what actually matter. Tools change fast, but the reasoning required to piece them together safely doesn't. When we stop trying to master every new release and focus on fundamental problem-solving, the anxiety drops significantly.
How are you and your team handling the gap between management expecting instant AI-driven speed and the reality of taking time to verify and maintain that code?
You've named the anchor — judgment and design don't churn like tools do. On the management gap: reframe "slow" as "not shipping bugs at 2am," and make verification time visible, because speed only counts if the code survives production.
Spot on reframing. Making verification visible is such a key piece, if management doesn't see the safety checks, they just assume the code wrote itself in five seconds. Appreciate the advice!
Thank you!
What I miss most about the "old way" is the feeling of accomplishment, of understanding things.
Nowadays it just feels like you´re starting skills... the whole thing of reading the ticket, fixing the bug or implementing the feature, as well as committing, pushing and opening the pull request is gone, it´s all automated away.
And what do you do with all the short gaps you now have between starting the workflow for a new ticket and testing the ticket which was previously done?
Role seemed to has shifted from being an engineer to being a product manager, who has a little more technical knowledge.
That shift from making to managing is exactly the loss nobody warned us about — the accomplishment lived in the doing, and automating the doing quietly automated the satisfaction too. And those short gaps are the strangest part: too small to rest in, too empty to feel productive, so you end up busy without ever feeling like you built something. The PM-with-technical-knowledge role is real, and for a lot of us it's not the job we fell in love with.
That´s so true... have you already found some kind of solution for that?
Man I felt obligated to share this post since it clicked with me. I mean rather than being reactive I believe being proactive can help us building a shared understanding around it so nobody feels pressured or insecure just because they think everyone else has gotten it right already and they are falling behind.
That's the exact antidote — proactively building shared understanding, so nobody's silently assuming everyone else has it figured out. That assumption is half the weight, and it breaks the moment one person says "me too" out loud. Thanks for sharing it.
Until now change in tech used to be "incremental", people used to get time to digest the nature of change and act on it.
With this AI trend, there is not enough for people to understand the change and adapt, so that is for sure causing LOT of stress. In addition the env/industry is not in so good shape, so yeah stress is real for sure. Thanks for putting the thoughts out!
Exactly — it's the collapse of digestion time that makes this different; change was always constant, but you used to get long enough to absorb it and adapt, and that breathing room is what AI took. Thanks for reading.
I am Javascript developer. I recognized the feeling when new JS frameworks are released every month. AI era ended them but made the feeling worse. New models, best practices, stacks every week or even days.
I survive by one rule: own an opinionated standard about how things are arranged and implemented. It's mine, forged through years of experience. No AI models and frameworks will dictate that. I tell my AI to create functions, but the architecture is determined by me. I let them improvise within their instrument, but I am still the conductor of the orchestra. This synergy for me brings the best out of me and my AI model. It's slower than letting AI do all the magic, but I can confidently say "I am accountable for this code".
"I let them improvise within their instrument, but I'm the conductor" is the whole healthy relationship in one line — you kept the architecture, which is exactly the judgment that doesn't expire when the frameworks do. And "I'm accountable for this code" is the part that quietly protects your sanity and your craft at once.
This really comes at the right time, and I believe everyone should read it and take a step back from the idea that we need to constantly keep up with AI.
Let me share a much simpler way to deal with this (which I applied since ChatGPT comes out): If something new is released, I make sure I know about it, understand what it can do, and then… I don’t use it until I actually have a reason to.
That's the sanest filter there is — know it exists, understand what it does, then let it sit until a real need pulls it off the shelf, instead of chasing every release for its own sake.
wow..im just starting to learn about this IT world and what I read here left me speechless.. is it really that bad and stressfull working in it because of ai?..what I got from this post is that maybe just work with the ai and keep calm since the ones creating the ai 'set the pace' ? .. im just trying to figure out how to break into this but skip the ''loop'' as you guys called it..
Don't let this scare you off — the loop is a trap you can see coming and step around, and starting fresh, you actually have the advantage of learning to work with AI from day one instead of grieving an old way. Build strong fundamentals, stay curious, protect your rest early, and you'll be fine. Welcome in. 🙂
Thanks for your reply, I thought about the fact that now I can easily ask AI when encountering a problem instead of doing the scrolling through forums how you guys used to do it..hard work, a lot of respect for you. Wish you all the best, thanks for taking the time to reply
That respect goes both ways — you're starting in a harder, faster world than we did, and choosing to actually understand things instead of just asking is exactly what'll set you apart. All the best, genuinely.
The "volume vs. pace" distinction is the real thing. I've tracked my own output for 18 months — lines shipped, APIs deployed, bugs fixed — and the numbers went UP while the feeling of falling behind got worse. That's how you know it's not a workload problem. You can't sprint out of something that doesn't have a finish line, and I think most of us are still trying.
Rising numbers plus rising dread is the clearest proof it's pace, not volume.
Finishing code used to mean being done; now I'm lying in bed wondering whether the module the agent rewrote this afternoon has a trap in it. Faster output means a heavier review tax, and nobody books that on any dashboard. Last month I made myself a rule: agent changes don't merge until I've actually reviewed them — if I run out of day, it waits. Output dipped the first week, but that background process finally shut down, and my sleep genuinely improved.
That rule bought back your sleep — best trade you'll make.
Maybe unpopular, but I see it almost the opposite way. AI is making tech move insanely fast, but it’s also never been easier to learn, experiment, and build. Keeping up with everything was never the job. Things were moving fast before AI as well.
Not unpopular at all — and you're right that "keep up with everything" was never the job, AI just cranked the volume on a pressure that always existed. The same speed that overwhelms is also the lowest barrier to learning and building we've ever had; which side you feel more depends a lot on where you're standing. Both are true, and it's a fair pushback.
This really resonates with the current tech environment. There’s always a new tool or model to learn, and it’s easy to feel like you’re falling behind. I liked the reminder that we don’t have to keep up with everything—building strong fundamentals and staying healthy matters too. 👍
Exactly — the fundamentals and your own wellbeing are the two things that actually compound, while the "keep up with everything" race was never winnable anyway. Thanks for reading.
Thank you for this.
I am (currently) 56 years old, I have been developing software since I was about 13 years old. More or less. I love what I do, I have written a book about why "writing the code is 10 percent of it" and I still think that using computers to solve real world problems is cool.
Now ... I have fallen between the planks. It's not "AI as such" that has, very suddenly, robbed my of my income. It's politics as well as a shift in the world's "vibe" - and it is the latter that I would like to add to your list:
In my experience, developing good software solutions has always had that component of "it actually really does what it needs to do", in order to get THAT quality, developers had to wear many hats, understand the disciplines they were working with, talk to and with people and listen, learn, understand, bring in new ideas from elsewhere and HELP PEOPLE do their jobs better. Put that in other words if you will, the point is: We did pride ourselves in KNOWING what we do and DOING A GOOD JOB.
That is gone. When I look at developers these days, they pride themselves in running 40 agents at the same time working on 10 or more projects in parallel and generating shipable products within days. Who uses those products? Aside from other agents developing more tools? Who is doing a BETTER JOB with something that exclusively outputs a once-and-for-all defined (prompt based) result? We get tools over tools over tools, we drown in "solutions" and get done less than before the PET2001 was a thing (that's my feeling, not reality, I know).
It's not just the pace and the ground. It is a general, overwhelming, everywhere "we don't care". It's "just use this". It's "ask chatGPT". It's "I don't have to learn it, that's what AI is for". It is the everywhere-nobody-gives-a-damned-shit vibe.
No matter how good you are. No matter how smart, fast, experienced, intelligent: It does not matter, you can't keep up anyway and even if you could, who cares?
I have started singing half a year ago. The progress I am making with using my voice as a musical instrument is giving me the kind of happiness I used to have with solving a tricky workflow problem for a client that made users call me and say a personal "thank you, what you did there makes our days a real bit easier". Over the last five years NOBODY thanked me for all the work I did in software. I think I am not just "out of job", I think I am out of "I want this job".
This is one of the most moving things anyone has left here, and I don't want to shrink it. You named a loss deeper than pace: the craft ethic itself — knowing what you do, doing it well, helping real people, and being thanked by a human voice. The "nobody-gives-a-damn" vibe you describe is real, and losing the part where the work mattered to someone is a grief the productivity charts will never show. That it's been five years since anyone said thank you — that's the actual wound, and it's a legitimate one. So maybe you're not out of the job. Maybe you're out of a version of it that stopped letting you do the thing you loved: make someone's day genuinely better and hear them say so. The singing makes complete sense — it gives back exactly what the work stopped giving. Forty-three years of caring about doing it right isn't erased; it just needs somewhere that still values it. Thank you for writing this. Truly.
"The most honest piece I've read in tech in years. The constant background anxiety of 'am I doing enough' has replaced the joy of actually building things. Thank you for saying it out loud—consider me setting it down, too."
That last line got me — "setting it down, too" is exactly what I hoped this might let a few people do. Thank you.
the 'vigilance that never turns off' is the one I keep returning to. it's not just that you review the output — it's that the model is confident about everything, including the things it got wrong, so you end up in a higher stakes reading mode than before the tool existed. you're not reading to understand, you're reading to catch. that's a different cognitive load and it doesn't feel like 'help.'
the identity wobble cuts differently in practice: I've noticed I'm less proud of work I'm genuinely proud of, because I can't always draw the line precisely. that's a new kind of weird.
what's actually helped you with the vigilance load, if anything?
"Reading to catch, not to understand" is the sharpest description of the new load I've seen — the model's flat confidence means you can never lower your guard, and that constant guarding is exhausting in a way "help" shouldn't be. What's helped me most: deciding up front which parts actually need the high-stakes read (anything touching data, money, auth) and letting the low-stakes stuff be low-stakes, so the vigilance is aimed instead of always-on. And on the pride wobble — I've started measuring it by "could I explain and defend every line?" rather than "did I type it," which redraws the ownership line somewhere I can actually stand.
After more than 20 years working in tech, I think AI has brought some amazing possibilities, but also a lot of pressure to constantly learn and deliver more. As a software architect, I use tools like Codex in my daily work, and while they help a lot, I also think we need to be careful not to confuse speed with quality. At the end of the day, we're still the ones who need to understand the problem, make decisions, and take responsibility for what we deliver. And with everything changing so fast, I think it's important to remember that we also need time to learn, make mistakes, and just catch our breath sometimes.
Twenty years in, you've earned the perspective that speed isn't quality — and it's the thing the hype keeps hoping we'll forget. The tools accelerate the output, but understanding the problem, making the call, and owning the result never got automated, and never will. And that last line matters most: needing time to learn, to make mistakes, to breathe isn't falling behind — it's the only sustainable way to stay good. Thank you for reading it.
The verification part is what got me. When I use AI heavily, the work shifts from writing to checking everything it touched. Faster output can still leave you more mentally cooked at the end of the day.
Exactly — the effort didn't vanish, it just moved from creating to checking, and checking someone else's confident output all day is a quieter but heavier kind of tired.
This is exactly what I’ve been feeling
😀
The "floor keeps rising" point is the one that hit me hardest. When "fast" becomes the baseline, you can ship more than ever and still end the week feeling behind, because nobody announced that "enough" moved. That explains why the tiredness doesn't match the workload.
I also agree that the silence is part of the pressure. Everyone performs "I love this, it's an exciting time" while privately running on fumes, and each performance makes the next person feel more alone. Your point that the comparison trap breaks the moment one person admits they're tired too is right. I've seen a single honest "me too" in a team chat change the whole mood.
The distinction between staying current and staying whole is a useful one. Frameworks turn over quickly, but judgment and taste hold their value. That's where I'd rather spend limited energy.
To answer your question: for me, the heaviest part is the vigilance tax, never being able to just trust the output and relax. It's a background process that never quits.
Thanks for saying it out loud.
The vigilance tax is the one I most wanted people to name, because it's the most invisible — it doesn't show up as work, it shows up as never fully resting, a background process that runs even when you've closed the laptop. And you put "the floor keeps rising" better than I did: nobody announces that "enough" moved, so you measure yourself against a line that quietly climbed, and the tiredness stops matching the workload because it isn't from the workload — it's from the chase. That single honest "me too" changing a whole team's mood is the most hopeful thing in your comment, and it's the cheapest intervention any of us has. Thanks for reading it whole.
This article captures the quiet exhaustion of AI-era work perfectly. We don’t need to chase every tool; protecting our judgment, fundamentals, health, and boundaries matters far more.
Exactly — protect the judgment and the person; the tools will keep changing regardless.
Real, its basically just imposter syndrome
I've done plenty of endless AI iterative sessions which have resulted in what I call "AI Hangovers"
"AI hangovers" — painfully accurate. 😅
Hi have yall been here a long time.
Your post hit me at just the right time. I have been feeling everything you expressed and was confused on what was wrong, but hearing your explanation makes me feel less like something is wrong with me, thank you. My biggest struggle other than feeling tired, was questioning if this is even a career I feel satisfied in anymore. I joke about this but I feel like I have become a glorified AI wrangler, and each day I feel like I get further away from why I started this journey. Your section on "Protect the parts of the work that are still yours." directly reflects this concern, helping me feel at ease in this constant state of anxiety. thank you again
This is an amazing comment section with really good insight into a problem I have suffered from as well. HOWEVER i find it odd that despite everybody feeling a certain way on how AI is impacting their life, they don't pick up on or question the author, comments being made in front of their faces with literal AI slop.
Don't get me wrong; there's genuine sparse genuine comments that spoke to me as a 10+ year architect/engineer.
But the author "James Anderson" is almost certainly using Claude in an automated content-farm workflow to reply and write their content. Analyzing their articles, it's clear this is nothing new. Repurposed content from other authors..
I mean, run it through Claude Opus yourself and you'll see (96% confidence AI-authored content using Claude).
AI RESPONSE BELOW
This is almost certainly an AI-managed author persona — likely running a prompt along the lines of:
The article itself reads identically to the replies. With 28 published articles all in the same register, all on AI/tech topics, this has the profile of a Claude-powered content farm persona on DEV.to.
Behavioral Tells
END AI ANALYSIS
I just felt like I would point this out. Everyone here is feeling the AI burnout; yet we refuse to detach from the tools (or maybe we're unable to anymore).
Sorry for being that guy but... let's get some human creativity back on dev.to eh? 🫠
Add to that the exhaustion of having to read AI written text all the time 😞
The best description of this issue I've seen was from a commentator using a call-centre worker as an analogy (I wish I could find it again to post a link; because it's so spot on): the automated chat-bots handle all the easy calls; which leaves the call centre worker only ever dealing with the more difficult issues. So there are no easy wins. No quick pleasant exchanges where you solve a problem for a happy and grateful customer. The people you're dealing with are already frustrated after wasting time with the chat-bot. There is simply no respite and only constant pressure...
The reality is that the push for greater efficiency and productivity benefits no-one except the people up the chain. Is anyone reducing staff hours because AI has made them more effective? Do you feel any deep satisfaction after confirming the AI fixed a bug for you? Are you actually expanding your knowledge and gaining better understanding of the code?
If we only measure "progress" by worker productivity and business profit - rather than job satisfaction and a quantifiable benefit for society - we are doomed to be slaves to a broken system.
I used to track all this A.I. progress and new tools on the daily, keeping up with models and approaches. I've learned a lot and it was fun, but I have decided to pull back a bit.
Constantly chasing changes for tools that will inevitably change tomorrow gets exhausting. So, if I miss a model release or miss a new framework, it will find me by way of necessity.
Like, if my project calls for whatever that feature or tool is offering, I will likely find everything I need during my discovery and design phases of the project. To some degree, it is like the overabundance of frameworks and tools in general we developers already know -- you don't focus on everything, you focus on what is relevant for what you are doing at the time.
The article captures an interesting side of working in tech during the AI era that is easy to overlook. Beyond productivity and new tools, there is also the uncertainty of constantly adapting to changing expectations, workflows, and the role of human judgment in software development.
This really resonated with me, especially the idea that the pressure isn't just about the amount of work anymore — it's the feeling that the ground keeps moving underneath us.
As a frontend developer currently expanding into backend development, I've definitely felt that pressure to keep learning new tools while also trying to build real-world experience. But I liked your point about separating “staying current” from “staying whole.”
I’m slowly learning that we don't need to chase every new tool. Strong fundamentals, good judgment, and the ability to keep learning probably matter much more in the long run.
Really appreciate the honesty in this post. It’s refreshing to see someone talk about the pressure behind the constant “AI is making everything faster” narrative.
❤️