I’m starting this topic as a place to discuss strategies for how to navigate professional software engineering in 2026, now that these AI “tools” have been so widely integrated into the profession.
Personally, I’m past the point of asking whether or not AI is good or bad. With the help of Brett and many others, I’ve realized AI is bad, full-stop, for all the reasons people here are familiar with. At the same time, I just started a new job and outside the AI stuff, it’s a pretty great opportunity that I am thankful for. Therefore, I’d love to hear people’s ideas for how to maximize this possibility of staying at this job in this career (or one related to it) while still avoiding at least personally using AI tools.
I’d love to hear people’s thoughts on:
Is it even viable to continue as a professional software dev in 2026 while pushing back against AI tooling and practices?
Is it better to proactively share anti-AI sentiment with your employer/boss/colleagues or should you “keep it to yourself” as much as possible?
If you do share it, what’s the best approach to communicating it? 1-1 with your boss? Post it on slack? Casually with your colleagues?
I know so much of this is dependent on where you work, your seniority level, who your boss is, etc. but I still think there are important lessons we could share that might help us when we face these similar challenges.
To start, I’ll share that I’m currently thinking of drafting like a brief document outlining my concerns about AI-driven development. While a lot of my concerns are extremely ethical and political, in this doc I’m planning on really emphasizing first that I think AI development is actually hurting us accomplishing our core missions in both the short and long-term. My worry is if I start the discussion with “AI is destroying the environment” (which it very much is) that makes me look ideological and “not a team player” if that makes any sense. Whereas the approach I think I’ll take will hopefully make me come off as caring too much, if anything.
Once I draft and hone the document, I’ll probably start with going over the points and then sharing it in a 1-1 with my boss. I know this is super me specific, but I have a pretty good relationship with my boss right now. He’s been in software engineering for 20+ years and while he doesn’t make it public, I imagine privately he shares at least some of the concerns expressed on this site.
After that, my main project manager is probably going to be the most delicate conversation. He’s a really nice guy and we get along, but he’s extremely AI-brained. It feels like so many times I’ll ask him a question and he’ll essentially say “ask the AI” when I feel like the premise of my question is that the problem is with AI! I expect I’ll share my concerns and doc with him the same way as I do with my boss. With him, I’m sure the main concern is going to be how do we actually do day-to-day work if I’m not using AI? What are my limits? Will I review PR’s written by AI? Am I okay with AI code reviews if I don’t write the code with AI? What happens if I go too slow? I can say here that ideally there would be no AI anywhere used by anyone, but I can’t force other people or the rest of the org to stop using these tools, unfortunately.
Once I write this doc I’ll probably post it to this forum for feedback. What do people think? Is there a better way to go about this?
I think when it comes to whether it is possible to maintain no AI it depends on a few factors
Speed of movement. If you are in a slower moving or more organised place managed by technical people than yeah it’s possible.
The Web dev or other dev split. If you are in systems programming the hold out camp is more vocal and larger. Even the pro-AI camp is more skeptical of code generation beyond autocomplete in languages like C/C++. Memory management and performance is not something AI does well and security concerns have not been mitigated with newer models.
Proximity to business. Ofc all paid programming has to make money to some extent - but if you and the business are in close coordination they do not care about code quality (except when it impacts them). Despite the engineering side trying every SDLC paradigm that one can think up the business side has given little care to the devs time, effort, or dignity. If they change requirements it is on the devs to make that happen in the same time frame. No other field of engineering accepts this as a given besides software.
Over time I see a backlash against AI software coming even from layman. Everyone complains about lack of software quality so I do see in the future commission/contract work for software being contingent on no AI and a boom of start ups from senior devs leaving. Those opportunities will likely not be the same pay or frequency as commercial.
I have found some success when it comes to communicating anti-AI sentiment to simply bring up neutral observations about code quality, bugs, or things AI can’t do well. My boss is pro-AI BUT we have no mandate (just tight deadlines making me rely on it) so YMMV. It becomes very hard to argue especially when there are times you can trace issues back to AI. Another thing is the copyright/ownership aspect since things generated with AI are not copyright-able.
My last workplace was pretty pro AI, not in any official policy but it was pretty unanimous across all teams that AI was the way to go for everything. When it came up (my manager asked why my token usage for the Copilot subscription they provided was 0) I just said I don’t use AI for my work. There was definitely some skepticism but it didn’t become an explicit issue.
I’ve generally stuck to not broadcasting my views on AI in the workplace unless explicitly asked, and that’s worked well so far. I’ve had tricky conversations with coworkers on why I don’t use AI, as well as asking pointed questions on the shortcomings other people have experienced when it comes up in team conversations. It can ruffle some feathers but I think it’s important to be the dissenting voice on the AI issue, otherwise it tends to get blanket approval to be used for any and everything.
Another part of that I’ve found helpful is to have alternatives to offer, even if its just reminding people how we used to do x or y 5 years ago. 5 years is an eternity in model time but it wasn’t that long ago and we were all there!
At my current place we are a bit of slow adopter to AI but there is strong push from nearest leadership that we should do more with AI.
I believe the main reasons are
Investors are pushing for it,
Other companies say do it so FOMO,
Anthropic et al hire the smartest people in the world (they say) so if they say it’s great so who are we to argue against it.
My strategy is non-confrontational currently, I don’t think I get people to listen right now, also selfishly I am getting a bit old and wonder if I lose this job will I find a new one?
WRT AI I have short and long term worries.
Short term I worry over that my coworkers rely on agents and agents then hijacks the credentials and do stupid things with it. Unfortunately I think we have to learn by having a serious incident as the mood is so Pro AI right now.
Long term worry is more of environmental impact, social impact and the destruction of capacity for thoughtful introspection.
I am not buying the AGI will destroy humanity stories the chief doom trollers spread but by using LLMSs to automate critical decisions I think we can manage to seriously hurt ourselves AGI or no AGI.
I’m in the “non-confrontational & casual discussions with colleagues” camp.
There are several approaches that might or might not work depending on the level of "pro-AI"ness of your interlocutors. One that’s been working quite well with managers is asking genuine questions about supposed productivity gains and ROI.
“What are your expectations about productivity ?” This paves the way to the whole “productivity in software engineering” discussion starting with “how do you measure it” and then “what makes you think using AI will help in anything ?” This was met with vague answers and “just do it” kind-of requests back in February, so I took on me to set up “DORA metrics” for our team, from June 2025 and up to March 2026 and then from April 2026 up to now. It showed using AI made us merge PRs more often in the beginning but with degraded quality which led to regressions and eventually slowdown. It might not work the same way for you. but I’d recommend you start tracking these metrics and show them to the team once you have some history.
With coworkers, I’d share what makes me hate my job right now. From coding (26+ years professionnally now) to basically “tell the bot what to do then check it didn’t do mistakes”. Like “oh you like baking cakes ? Well just put the ingredients into this machine and check out the cake is good ; if not, fix the instructions for the machine. Congrats, that’s your job now”. Basically we train the bot and the company’s processes how to replace us, while making stuff we loathe.
I wouldn’t share my sentiment to my boss, at least not in a direct, formal way. If you have a good relationship and now how to put the subject on the table, you might try that at an informal occasion. Although I could see a 1-1 working in a good organization.
If you’re really into it, you can have a debate about how intenable the economics around AI are. It just cannot work. But you better have some solid material to back that up ; personnally I like reading Ed Zitron’s pieces for a variety of reasons but it’s quite a mouthful (30-60 minute-long articles)
Good luck, it really is a horrible era for many people, not only in dev or software engineering.
I started writing a ton about this in a big old response but I think this is going to be an ongoing topic that is really contextual based around where one works and the environment, goals, and culture. I’ve got a lot of thoughts on this, from how to organize against AI internally to how to try to rein it in at a workplace where it’s being used haphazardly.
But I think the best place to start when talking about this is the erosion of collaboration that occurs when a team of software developers leans hard into AI. This is something that managers and business leaders are likely not aware of from their positions and definitely do not want. This is something that happens naturally when you have a chatbot you can ask questions to. Junior engineers and people new to the team don’t ask their colleagues questions because they can just ask the chatbot. When people are stuck, they don’t ask for help from a human, they ask the chatbot. The problem with this is that the human relationships weaken and collaboration atrophies.
Making good software requires collaboration and trust. That’s how knowledge is shared. That’s how problem solving on a team works. In a team that’s using AI, it now requires being more intentional about mentorship and problem-solving together. Teams should have policies against encouraging jr members and new hires to not ask the chatbot.
The best software teams are those that are collaborative, trusting, supporting one another. Leaning heavily into AI erodes that and creates knowledge silos. These are all things managers and business owners do not want. So this can be an effective angle to discuss with management and leadership, specifically with the goal/aim of increasing collaboration, team-building, and trust, not eroding it.
If an employer is measuring “AI adoption” in terms of token spend, one compromise is to write the code yourself and use the model for “feedback”.
Even if the feedback is wrong or useless, you can say “I used X amount of tokens and the AI said my code was great” where X puts you in the middle of the expected token budget.
On copyright:
I am worried about this angle because I don’t think the companies view it that way. For an example: the artists at my job now have to use a tool called Layer.ai to generate images. Layer advertises itself as a service where you can legally and commercially own the output (from this page):
Commercial Use
Information on commercial use and all things legal
December 30, 2025
Do I own the rights to assets that I produce using Layer?
Yes. You own the assets you create with Layer, provided you comply with our Terms of Service and Acceptable Use Policy.
Can I use Layer-generated images for commercial projects?
Yes. You are free to use the images you generate for any commercial purpose, including games, marketing, and product design.
What can I provide to platforms (e.g., Google) if they ask for image rights or permissions?
You can send confirmation that the image was AI-generated using Layer and that you have the rights to use it commercially.
Between Layer and our users, our users own the copyright to their artwork, what they create and any derivative work on the platform also belongs to them. Secondly, starter styles which are specified as such by Layer, are all owned or licensed by us.
I haaaate these blurbs because they’re BS, and ignore the fact that the underlying image generation tech that Layer is built on top of ABSOLUTELY stole a bunch of work, including copyrighted work, into its dataset, which is evident from the fact that mickey mouse keeps showing up in the output. But my company believes it’s legally sound, so folks there haven’t been swayed by any arguments about copyright, and they don’t expect to be sued. : (
I know that’s not code but I imagine other companies that provide code services are probably saying analogous things? : (
EDIT: don’t want to end this on too negative a note, though, maybe would say i’m just focusing on other arguments right now instead of this one at work. I’m getting a sense maybe my leads are more receptive to the angle of “I don’t like that we’re seeing mass layoffs and institutional skill deterioration”.
Regarding Layers claims, the first part is legit. You are able to use, sell, and have ownership over AI generated content. If the only concern for a company is that AI content can be sold and give no concern to IP than yes. Are there certain use cases where it doesn’t matter (a button here and there, some boiler text) sure
The questions comes in when something is FULLY AI generated. Idk how layers works BUT if the content is produced entirely with an LLM, prompts are not copyright able or enough to express intent:
The mere selection of prompts, even if those prompts are detailed and are the product of some human effort, does not itself yield a copyrightable work, although this determination could change as technology evolves.
That last bit can’t override this.
It can say it does, and your company can believe it, but it cannot assert that in a court unless you actually touched the assets manually. Even then:
Where a work includes both human and AI-generated content, only the human contributions are potentially copyrightable.
A lot of greenfield modules and projects in coding land are AI generated mostly, so this will come to bite them. Code has long been ruled to be an expressive element in US copyright law.
Now this is where things get murky:
mostly AI generated is still protected provided it was changed in an iterative fashion with human input. What counts as iterative enough to turn it from a draft to a product is still to be seen.
How can the US copyright office, most of whom are not coders, going to discern what is AI code? AI writing has a feel to it but it’s a feel we have pattern matched by exposure. Code is much the same but without living and breathing that world right now there isn’t a good way to know. Though Claude and Gemini watermarking texts is a start.
Most companies have not gone “pencils down” as it is now called - but given the murkiness it is a cause for concern for management even if they pretend it is not.
Appreciate this detailed response! I think i did see a defcon talk done by an EFF leader about how in the early days of the internet they fought for code to be judged as expressive/creative speech, and there was something about how that had big copyright implications?
I’ll probably pull the copyright talk into a separate thread to not derail too hard into one specific thing. But, awesome if there IS stronger legal ground to fight on, and if the challenges are more about identifying violations. def would love to learn more about that.
(Edit: will apologize for my niche thread swerve by repasting the original questions haha
Is it even viable to continue as a professional software dev in 2026 while pushing back against AI tooling and practices?
Is it better to proactively share anti-AI sentiment with your employer/boss/colleagues or should you “keep it to yourself” as much as possible?
If you do share it, what’s the best approach to communicating it? 1-1 with your boss? Post it on slack? Casually with your colleagues?
This topic got me thinking a lot about ways to talk about the risks of using AI on a dev team with management, leadership, etc. I came up with 35 real risks and challenges I experienced first-hand:
Possibly useful approaches/topics/mindsets that might resonate with a given manager or boss or CTO?
I have found that the environmental impact argument against using AI services to not be effective in a business environment. Businesses essentially exist to exploit labor and the environment to make money. It’s too abstracted away, like a lot of environmental harms from technologies. People can read a research paper saying every prompt uses (X) ML of clean water to run it but they don’t then associate that with their daily usage and the problems implied by it.
I have also found the shaky copyright and usage of stolen data to train the models argument to be not too convincing in a workplace either.
But hopefully these challenges and risks feel grounded and from experience, not from speculation.
