I recently saw a post on Reddit about software developers and mental health in the age of AI. The issues resonated with me, and the pain described is very real:
I work for a small consulting company, we have maybe 15 employees, working essentially remotely on programming or infrastructure tasks. I'm responsible for organising this team event where the main reason is to share and harmonize as much as possible our AI workflows.
However, I want to hack it to speak a bit about mental health.
I believe we have an important mental health crisis coming. I've seen 2 colleagues leave the company because of it—at least in part—and I'm certainly feeling it myself. We're overwhelmed, we are confused, we have imposter syndrome because it's not the job we learned or signed-up for. We lose human connection, and we increase context switching.
It goes further than [this] but I think a lot of folks will understand this very well. So I'd like your opinion. I welcome structured feedback, general feeling, personal stories... I want to understand the landscape and do good for my team.
Whether or not we talk about it, AI is having an effect on us. Yes, there are documented issues like "AI Psychosis" that are getting a lot of attention right now. But those extreme issues aren't always the ones that cause the most problems for us on a daily basis.
Understanding our fear
I've been using AI tooling to write code for not quite 6 months now. At first, it was exciting! The speed at which I could build, debug and ship was incredible. But as I moved forward and began to build out new and better ways of working with the AI tooling, I found myself running head-first into fear. Fear of losing my job. Fear of losing my own sense of self and the value I add to a team. Fear of what might happen to me in the near future, as more and more of the world begins to rely on AI.
I've mentioned this a number of times to my coworkers here at Test Double. In a recent conversation, Mike Doel said something that has stuck with me:
It makes me think of Atomic Habits by James Clear. He talks about three levels of change: outcome-based habits, process-based habits, and identity-based habits. The central thesis of his book is that habit formation is most effective when it is grounded in identity-based habits. How we see ourselves is so very powerful on what we do and how we feel about it. For all of us who cultivated an identity of being a programmer for decades, this moment that is challenging all of that is difficult precisely because it is asking us to reassess our identity.
This hit home for me and the fear I've been feeling - especially that last sentence:
For all of us who cultivated an identity of being a programmer for decades, this moment that is challenging all of that is difficult precisely because it is asking us to reassess our identity.
Identity is the hard part
Anyone who knows me outside of work can tell you that I'm no stranger to this kind of fear. I've reassessed, reevaluated, and rebuilt my own understanding of who I am, what I believe, and how I interact in this world, a countless number of times over the last twenty years. And what this person is describing on Reddit is exactly what we are collectively going through right now, and what I have done so many times before.
But having done this before doesn't mean it's easy for me. It's still incredibly painful. It's still causing a high level of anxiety for me. It's still a significant source of fear. And the uncomfortable truth about all of this—having lived through it many times already—is that the only way out is to push through.
We have to sit with the pain, discomfort, and fear. We need to understand why it's there, what it means, and how to move forward with our lives in this new world. Nobody gets to skip that part. And a person who has never had to question their own beliefs and values before is going to be in serious danger trying to navigate this alone.
Where to start
I wish I had the magic words to say, right here, that would make this easy for everyone to work through. Unfortunately, there are no such magic words. I do, however, have some advice on how I have worked through this in the past, and how I'm working through it now, to understand where I fit in the AI flood.
1. Therapy
I've seen a therapist for many reasons in the last 10 or 15 years. It's difficult, painful, and horribly stigmatized in our culture. But it's an absolute must for anyone who's experiencing major life changes, in my opinion.
My first piece of advice, then, is to make sure your company provides resources for each person to find a therapist. There are many different therapists working from various frameworks and perspectives. This matters. Different people need different therapists to match their specific worldview and needs.
Therapy must never be mandated. It should be available, it should be encouraged, and it should be easy to start. But it should not be required of anyone.
2. Do one thing at a time
Beyond the mental health expertise, advice, and assistance of a therapist, I've personally found that massive shifts in identity, tooling, responsibilities, etc, like we're now seeing, must be taken slowly. If we don't, we risk burning out or worse.
If you try to answer every question, every concern, every issue, every unknown, all in one gathering, you are going to run into analysis paralysis. The infinite choices and options become overwhelming, which easily leads to inaction and further stagnation.
Take things one step at a time. Start small. Don't expect the change to happen overnight. Get your team comfortable using AI tooling for one specific thing first, and let that win settle in before reaching for the next one.
3. Start here
With "one thing at a time" in mind, my recommendation for where to start with AI tooling is documentation maintenance. It's often the most innocuous place to begin because it's work most developers actively dislike doing.
However, your team, company, project, etc, may have people who actively maintain the documentation already. If this is true, don't yank the work out from under them. Find the places where your team would rather not be spending their time, and point the new tooling at those tasks. That's where you can start simple and get honest buy-in without anyone feeling like they are being pushed off a cliff.
4. Pick a tool, any tool
The question of where to start is often overshadowed by the question of "how" to start. There are countless options, choices to be made, tools to evaluate, processes to work with, plugins, skills, blog posts, articles, and so on. It's easy to feel like you're going to drown in the sea of AI opinions and options.
The truth of this situation is simple, though: the specific agent, tool, or framework doesn't matter for the first efforts. Pick one and try it out. You don't need a reason. You don't need an analysis document ending in a single, final recommendation. You only need to start.
What matters more is the goal—the "one thing" that will be used as a starting point, like updating old documentation. Let someone pick a tool to try because they said they want to try it, and see what happens. As long as they have a goal in mind, the tools and options will quickly narrow through trial and error.
5. Treat it like R&D
Lastly, treat this like any R&D project. It's learning. It's painful. It's scary. The team will make mistakes. Sometimes those mistakes will cause financial loss through time, effort, errors, and bad results. This is to be expected, and must be accounted for in the project of learning AI tooling.
Like any other project, the team needs to do retros and collect lessons learned and unanswered questions. The team must examine how people really feel about the work and the tooling. And they must feel safe in the environment and company culture to offer honest feedback.
Moving forward, carefully
Nobody has the full picture of what software development looks like in this new world yet. The honest move, as a leader walking a team through it, is not to pretend you do. Move deliberately, pay attention to what works, and let your team watch you do the same. People trust honest uncertainty far more than they trust a confident answer that falls apart a month later.
Our job, as the people trying to guide this, is to help each other figure out how we fit into the work going forward. Small steps, honest retros, and real support for the humans doing the work are critical. That's how we get through this together.
River Lynn Bailey is a Senior Software Consultant at Test Double, and has experience in building agentic workflows and plugins for solo and small team environments.
