I hesitated to write this.
Not because I don’t have an opinion about AI in software engineering, but because it sometimes feels increasingly difficult to have nuanced conversations about it.
I worked in an environment where being “AI native” was part of the identity. And when a technology becomes part of a company’s identity, questioning it can feel almost uncomfortable. The conversation quickly moves from “where does this create value?” to “how do we put AI everywhere?”
But I think we need more honest conversations.
AI is powerful. There is no doubt about that.
It helps me write code faster. It helps me explore unfamiliar areas of a codebase. It can remove repetitive work and accelerate experimentation.
But there is something important we should not forget:
Speed is not the same as progress.
Not everything needs AI
The truth is, not every problem needs artificial intelligence.
Not every business needs AI. Not every product needs AI features. Not every interaction needs to be optimized or automated.
Especially in industries built around people.
Hospitality, for example, is fundamentally about human connection. The experience someone remembers is often not the automated process behind the scenes, it is the person who cared, noticed, helped, and went the extra mile.
There is a Greek word I love: meraki.
It means putting a part of yourself into what you do. Doing something with love, creativity, and care.
And I think this concept is becoming increasingly important in software engineering.
The craft behind the code
When you genuinely care about your work, you don’t just want to eliminate every inconvenient part of it.
The difficult parts are where you grow.
Reading confusing code. Debugging strange production issues. Getting feedback from teammates. Having discussions about architecture. Making mistakes and learning from them.
These experiences build engineering intuition.
Software engineering has changed dramatically in the last few years. AI-assisted development is becoming part of our daily workflow, and I believe there is a lot of value in it.
But the reason AI makes me better today is because I learned the fundamentals before it existed.
I learned how systems behave.
I learned how to debug without an answer being generated for me.
I learned how to question solutions instead of blindly accepting them.
I learned through feedback, collaboration, and years of solving problems.
AI amplifies existing knowledge. It does not magically create engineering judgment.
What happens when we outsource thinking?
One of the things I worry about is not AI itself.
It is the temptation to outsource the parts of engineering that make us engineers.
Architecture exploration.
Technical discussions.
Understanding trade-offs.
Questioning assumptions.
Designing solutions together.
A company is not just a machine that transforms requirements into code.
A company is a collection of people with different experiences, perspectives, and expertise. The best solutions usually appear through conversations, disagreements, and collaboration.
If we delegate all exploration and architecture decisions to agents, what happens to the collective intelligence of the team?
What happens to the engineers who no longer build intuition because they never had to struggle?
What happens to the products created by people who never deeply understood the problems they were solving?
We might become very efficient at producing software nobody feels connected to.
Building software with meaning
I miss the feeling of building something meaningful.
I miss challenging architecture decisions. I miss deep technical discussions. I miss the excitement of solving hard problems with people who care.
Because the best software is not created by the fastest person typing the most code.
It is created by people who understand the problem, care about the users, and bring their experience into every decision.
I love building useful products.
I love seeing customers happy because something we created solved a real problem.
I love software engineering because it is a craft.
AI can help us create faster. It can help us explore more. It can remove friction.
But we should be careful not to remove the very things that make building software meaningful.
The future of engineering should not be humans versus AI.
It should be humans using AI while protecting curiosity, craftsmanship, and ownership.
Because without those things, we are not building software. We are just generating output.
Epilogue
Thank you for reading my messy thoughts, since today i got laid off because my company wants to be AI native, it felt appropriate to share my fears for the future to a bunch of strangers.
Top comments (26)
Hey Alexandra! Hope you are well and welcome back on DEV!
I am so sorry that you got laid off due to AI. It has become an unfortunate thing nowadays because of how AI is seen as a "productive" boost, but in reality, it hurts them slowly.
I believe companies are slowly realizing this based on public opinion. When looking at Microsoft and how they add "AI" everywhere on their services and products, people are slowly turning away from using Microsoft Products, even GitHub surprisingly because of how unorganized it has become.
I think in the near future that AI is not gonna go away anytime soon. However, we will start to see a more focus view on AI and when it should be used the right way.
I wish you the best of luck on your journey as a dev. Keep us posted here, yes? We would love to hear your experiences so far and I think others will see your value and learn from you! :D
Hey, thanks for your reply!
In the last months, I built skills, I automated, I adopted all the ways my company wanted to adopt. And just yesterday I wrote a report that AI didn't helped us at all to customer cases. Isn't that poetic? 9/10 of the cases were solved by a human, so what logical step to take? Fire them!
If we don't build products for our customers, who are we building them for?
I am curious on how long you worked for the company for and if you enjoyed there? Might get some better insights because it sounds like the company isn't worthy to be part of.
I was working there for two years, it started fun and exciting. This shift happened since January with over spending in tokens since May. There was no additional revenue to justify spending millions on tokens, and I guess they had to break even. Note: I was expecting layoffs but not in this scale (around 35% of rnd)
Oh geez. That's rough...
What are your current plans at the moment?
My plans is that I have no plans, I want to stay still a bit, remove the AI co-dependence they were trying to implant in my brain, and start love coding again
And I guess leetcode? Llmcode?
I wasn't expecting a turf war over profile descriptions. 😄
I have to admit, "I use a lot of Ygritte gifs to sass you into programming challenges, Jon." is hard to compete with.
And thank you for the kind words. Yesterday was a rough day, but you're right: work isn't the whole story. Companies come and go.
My bio is a legacy from when I first joined, when coding was fun and frontend not a chore, it would be a pity to change it. Back then I was making digital content and it sounded catchy to my non-native English brain!
Turf war is already in progress if you check your GH issues.
Even Peach Kelli Pop has been deployed.
I am sorry you got laid off, I too feel that change. Moved from regular employee to contract work. It feels strange to not feel like I am worth the money and support.
However, I am already seeing issues arise from outsourcing front end work. From a business stand, I think there are people who want to find the shortcuts and others who see coders as long term assets of the company. I hope you find something better and worth your efforts and knowledge.
@ale3oula it sucks but I know you'll bounce back! The job market will always repair itself and when that time comes, we must be ready to take new opportunities.
The point about AI amplifying existing knowledge rather than creating engineering judgement is important. Used properly, AI can remove repetitive work and help engineers move faster through automation, but architecture, trade-offs, domain understanding and responsibility still require experienced people. Replacing that expertise in the name of efficiency is likely to create more problems than it solves. As with most things involving AI, I believe the answer will be a hybrid, human-in-the-loop model, where AI handles the tasks it is best suited to and people retain responsibility for the areas that require judgement, experience and accountability.
The engineering skill that matters more in the AI era is probably judgment under abundance. Code is cheaper to produce, so deciding what should exist becomes more valuable. The best engineers will not be the ones who type the most; they will be the ones who can frame the problem, constrain the system, and verify the result.
Ngl, it's scary... I can smell it in my future too. Some context, my day-job is as a .Net dev, before this year, I was the only one using AI, now the boss got everyone packages for claude, I dont even use mine. But I can smell the layoffs in the future and I'm probably the first on that list, because I work from home. They asked me earlier the year to start driving to the office, under pretense that there's a missing 'synergy' that cant be gained from a rustdesk station, or a video call... Translation, they want me there, so I can teach them how to effectively use AI and they want access to my trained local model that lets me outperform the lot of them. Not because they want to boost production, but because they want to make me obsolete... The app is nearing completion and while maintenance is never truly complete, they also dont need 10 developers for it. So why keep the 1 guy who works remotely, even if he's fast? By chance, a week after they said I have to come to the office, I herniated my lower spine, so I got medically booked off to work from home for 3 months, it's now month 2... Without going into too much detail, driving an hour to work every day is not an option, I literally cant afford it, so I know they'll sack me sooner rather than later. The real problem is living smack dot in the middle of nowhere means no job market at all... Europe hires europeans, US hires US, India hires Indians, China hires Chinese... Who hires Namibians? Our currency isnt even listed... I've built some really cool stuff, I work hard and I work fast, but effectively, I am unhireable due to locality. That's a scary thought when you know you're getting sacked soon... Especially mid-process of buying a house, with a fiance and 6 pets as a single earner household. I'm scared honestly, it's actually why I originally joined on Dev.to. Out of desperation to get hired, I released V.A.L.I.D. and MCP-Lite as AGPL-V3, came here and posted about it, just so I can hopefully get word out that I'm a good developer and that someone would be interested in hiring me. The whole AI movement is frightening, but I fully understand why it's happening and I honestly cant blame them, it's just more efficient. but what happens to all of us on the chopping block when the job market shrinks by 30% along with 30% layoffs? "start your own company, build your own app", doesnt quite have the financial security a job does until you get it off the ground... It's absolutely terrifying and I honestly dont know how it's going to end.
The "meraki" concept really resonates. I've noticed the same thing — when AI handles all the friction, you lose the friction that actually builds skill.
One pattern I've found useful: use AI for exploration (understanding a new codebase, trying different approaches) but do the actual implementation by hand. The speed gain from AI exploration plus the learning gain from manual implementation seems to be a good sweet spot.
Speed without understanding is just faster accumulation of technical debt.
100% with you - I think AI can and should be used selectively, and where it adds the greated value - not "slapped" onto everything and anything "just because we can" ...
It's a tool, not the holy grail - companies which go overboard with it will find out the drawbacks of that approach the hard way ;-)
The engineer role is moving toward judgment under leverage. Writing code is still important, but the scarce skill is knowing what should be automated, what must be verified, and where a fast generated answer creates hidden maintenance debt. AI raises the floor, but it also raises the cost of weak review.
I agree with the core warning, but I think there's a risk of romanticizing struggle a bit too much. Not every difficult part of engineering is formative in a useful way. Some friction is genuinely educational, but some of it is just waste created by poor tooling, bad documentation, accidental complexity, or organizational dysfunction. If AI can remove the second category, that may actually create more room for the kind of meaningful craft you're seeking to preserve.
For example, reading confusing code can build intuition, but it can also simply burn hours on naming inconsistencies, missing context, and historical baggage that nobody intended to preserve as a learning exercise. In those cases, having AI summarize module boundaries, trace call paths, or explain likely intent can help engineers get to the real problem faster.
Where I strongly agree with you is that the line gets crossed when AI stops being a tool for reducing accidental complexity and starts replacing engagement with that complexity. Architecture trade-offs, domain understanding, user experience, and explicit gates and guardrails are not chores to delegate away. They are the real human work.
The thing is that if a junior developer, won't pass the struggle stage, (s)he will never become an engineer with the judgement that is build through that experiences. Even if it is try to debug a mess, retrospectively enhances our skills, even though we don't like it at that moment.
I agree on what you say, our job is not to write code per se, but to solve problems, but I also believe that we should use AI as a tool to applify our work and not delegate to something else the work - who cannot even own it.
Some comments may only be visible to logged-in visitors. Sign in to view all comments.