My workplace now expects us to use AI for essentially all coding unless we are willing to openly disregard management, and I'm worried that refusing could cost me my job. I'm much faster with tools like Claude, and they have genuinely helped me learn and solve problems, but generating code this way feels repetitive and has taken away much of the satisfaction I used to get from programming. I still enjoy architecture and planning, and I'm looking for a new role where I can have more hands-on involvement, although it feels irrational to intentionally work slower by avoiding a tool that makes me more productive.
I also deal with moral scrupulosity and OCD related to social judgment, which makes the situation harder. My personal social circle is very opposed to AI, and I avoid using it outside work because I want to make ethical choices. Recently, someone I know made an extreme comment about wanting to hurt people who use AI at work, and now I'm anxious about seeing her soon. I feel like I'm hiding an important part of my life from my friends and worry that using AI makes me hypocritical or morally compromised.
For those dealing with similar workplace pressure or ethical conflict, how do you think about it? How do you balance earning a living, using a tool that is genuinely useful, reviewing the code responsibly, and still feeling aligned with your values?
5 Answers
If the work itself has become joyless, that’s a valid reason to look for another role—but don’t assume you need to abandon software entirely. Some teams still value careful review, regulated development, infrastructure, systems design, accessibility, security, or other work where judgment and domain knowledge matter. In the meantime, use personal projects or manual coding practice to preserve the part of programming you actually enjoy, rather than expecting your job to provide all of it.
There’s no need to feel ashamed of using a productivity tool, but it’s worth protecting the skills that let you judge its output. Read the generated code, test it, understand the tradeoffs, and keep practicing architecture, debugging, and manual coding when you can. AI is most useful as an assistant or fast typist; it becomes dangerous when nobody involved understands the result.
If your employer requires the tool, using it is not the same as enthusiastically endorsing every company behind it or every way AI is deployed. You’re trying to keep your job, and you can still minimize harm by using the least expensive model that works, avoiding unnecessary prompts, and reviewing everything before it ships. Being responsible for the final code matters much more than whether you typed every character yourself.
The important distinction is whether you’re actually delivering reviewed, maintainable software or just accepting whatever the model produces. Faster code generation does not automatically mean faster or better delivery.
Ethical consumer choices are almost never perfectly consistent. You can object to the environmental and labor effects of AI while still using an employer-mandated system, just as people may oppose certain cloud providers without being able to replace their company’s infrastructure. You don’t owe everyone a confession or a detailed defense, but if you want to tell your friends, you can simply say that your job requires it, you’re uncomfortable with parts of it, and you’re looking for alternatives.
The OCD angle deserves care here. Your friends’ reactions are opinions, not evidence that you’ve done something immoral, and trying to achieve perfect moral purity can keep the anxiety going. Keep working with your therapist, notice the urge to seek reassurance or confess, and let the discomfort exist without treating it as a command. Also, someone joking about murdering people over a workplace tool is not a reasonable standard for judging you; you’re allowed to set boundaries around that kind of talk.
You can still take the environmental and social concerns seriously without making yourself personally responsible for solving an entire industry’s problems. Focus on the choices you genuinely control.
That’s especially important when non-developers are generating code. Without enough technical knowledge to ask the right questions or spot problems, the apparent speed can turn into slower delivery and a pile of maintenance issues.