[Draft] We Can't Automate Our Way Out of Burnout
Why mental health, trust, and team culture belong in tech's AI strategy, not on its list of things to deprioritize.
There is a message I keep hearing in tech: move faster. Use AI. Deliver more. Don’t let the company fall behind in the AI arms race.
I understand the urgency. I also keep wondering what happens when urgency stops being a response to a particular challenge and becomes the permanent culture.
Somewhere in that shift, the things that make work feel human can start to look expendable: a friendly greeting, time to get to know a teammate, a team lunch, room to ask for help, a workday that actually ends.
Nobody has to announce that these things no longer matter. It can be enough to make them feel like time we cannot afford.
My concern is not that we are taking AI too seriously. It is that we risk taking the people building with it less seriously.
[Opening example: Add a moment that made you notice this shift. Keep it grounded in your experience and leave out confidential details or identifiable colleagues.]
Culture is not a nice-to-have for everyone
I have a higher baseline of neuroticism, and culture is something I need at work. It was part of why I chose Microsoft.
[Personal context: What does a supportive culture look like to you? Add one or two examples: a manager who welcomes questions, teammates who check in, or the ability to admit you are stuck without feeling that your job is on the line.]
That is not the opposite of ambition. Wanting to do excellent work and needing a healthy environment in which to do it are not contradictory.
When I say culture, I do not just mean social events. I mean how people treat each other when deadlines are tight, whether expectations are clear, whether help is available, and whether it is safe to say that something is not working.
I am writing from my own perspective, not on behalf of my employer.
Constant change has a human cost
Layoffs, performance improvement plans (PIPs), and reorganizations are different things, but each can introduce uncertainty. A reorg can also break up relationships and routines that took time to build.
When those changes keep arriving, I worry that people can move from a collaboration mindset into a scarcity mindset:
- Is it safe to admit that I need help?
- Will sharing credit make my own contribution less visible?
- Should I spend time helping a teammate, or protecting my own position?
- Is taking time off going to look like a lack of commitment?
These are possible responses, not a description of every employee or every team. But they are questions worth asking before treating an anxious workplace as a high-performance one.
[Example to add: Describe how uncertainty can affect collaboration or your own ability to focus, without speculating about someone else’s mental health.]
Who are we designing this pace for?
Sometimes the ideal employee in this race seems to be an imaginary person: always available, endlessly adaptable, unaffected by uncertainty, and able to absorb more work without needing more recovery.
That is a narrow definition of who gets to thrive.
This is not about dividing people into “emotional” and “non-emotional,” or assuming that a neurotype determines someone’s ability to succeed. Everyone has emotions. People differ in temperament, support needs, responsibilities outside work, and how they respond to stress.
We should not confuse being good at tolerating a particular environment with being the only kind of person capable of excellent engineering.
If we want a range of people and perspectives building AI, we need working conditions in which that range of people can succeed.
Burnout is not a sustainable productivity strategy
I worry that some of what we celebrate as acceleration may be borrowing from people’s future capacity to work.
The World Health Organization’s overview of mental health at work identifies excessive workloads, low control, limited support, and job insecurity as risks to mental health. It also explains that healthier working environments can support retention, performance, and productivity.
That does not mean someone experiencing anxiety or illness is inherently less capable. It means working conditions can make it harder for people to do their best work, and organizations have a role in improving those conditions.
[Evidence to add: A recent, credible source on burnout trends in tech. Specify the population and time period; do not assume general mental-health statistics prove that AI is causing burnout to rise.]
If the short-term gain is more output, but the longer-term result is exhaustion, absence, or people leaving, are we measuring the full cost of that speed?
And even if there were no productivity benefit, people would still deserve a healthy workplace. The business case matters, but it should not be the only case.
Our mental health does not reset when we log in
Work is not happening in a vacuum. People arrive with lives, responsibilities, and stresses that do not disappear when they open their laptops.
I am currently reading Permission to Feel by Marc Brackett. The book’s publisher description frames emotional well-being as a concern for both children and adults, and introduces emotional skills as something we can learn rather than something we either naturally have or lack.
[Book passage placeholder: Choose a passage from your copy about wider emotional distress or the consequences of suppressing emotions. Add the edition and page number so any quotation can be checked before publication.]
[Your reflection: What has the book helped you notice about how emotions are handled at work? Separate Brackett’s argument from your own interpretation.]
For broader context, WHO estimates that depression and anxiety account for 12 billion lost working days annually worldwide. That estimate covers the wider economy; it is not a measure of tech burnout or evidence of an AI-specific effect.
The question I want to ask is what happens when workplace pressure compounds the stress people are already carrying, rather than giving them a stable place to contribute.
Making room for people is part of the strategy
I am not arguing against accountability, ambitious goals, or learning new tools. I am arguing against treating basic human needs as obstacles to those goals.
Some places to start:
- Make priorities explicit. If something new becomes urgent, say what can wait instead of quietly adding it to an already full workload.
- Protect recovery time. Do not make constant availability the unofficial requirement for being seen as committed.
- Preserve connection. Leave room for greetings, check-ins, and getting to know the people whose judgment and help you rely on.
- Make connection inclusive. Team bonding should not require after-hours attendance, extroversion, or sharing personal information.
- Communicate through change. Be clear about what is known, what is not, and when people can expect another update.
- Reward collaboration. Recognize mentoring, knowledge sharing, and helping teammates, not just the most visible individual output.
- Make it safe to raise a concern. Asking for help or questioning an unrealistic deadline should not be treated as a character flaw.
A team lunch cannot compensate for an impossible workload. Culture has to show up in decisions, not just activities.
[Practical example: Add one change you would like to see, or one practice that has helped you feel supported while still doing challenging work.]
Recommended reads / listens
Read: Permission to Feel by Marc Brackett
This is my current read, not a finished-book review. Brackett’s focus on learning to understand and work with emotions feels relevant to the question of what we expect people to leave at the door when they come to work.
[Reading note: Add an idea that has stayed with you, why it matters to this article, and a page reference if you discuss a specific passage.]
Watch / listen: Brene Brown, The Power of Vulnerability
Watch on YouTube | Official talk page
Brown explores vulnerability and human connection. My connection to engineering is this: admitting uncertainty, asking for help, and sharing an unfinished idea all require a willingness to be seen before we have everything figured out.
[Viewing note: Add the moment that resonated with you and a timestamp.]
Watch / listen: Brene Brown, Listening to Shame
Watch on YouTube | TED talk page
Brown examines shame and what happens when people confront it. The workplace question this raises for me is whether our teams make it easier to acknowledge mistakes and learn, or make people feel that a mistake defines their worth.
These talks are a lens for thinking about trust and collaboration, not direct evidence about productivity in AI teams. Vulnerability also should not mean pressuring colleagues to disclose private experiences.
[Viewing note: Add your own takeaway about how openness can support a more effective team.]
The race is not an excuse to forget the people running it
We can be ambitious about AI without accepting that the price of admission is exhaustion, emotional suppression, or a life organized entirely around work.
When we talk about human capital, I want us to remember the human part.
The goal should not be to find the few people who can endure an unhealthy pace. It should be to build an environment where different people can do excellent work sustainably.
[Closing thought: What do you want an individual teammate, a manager, or a senior leader to do differently after reading this?]

About the author
Michelle Mayer
A creative, people-minded software engineer exploring AI safety, ethics, mental health, and life beyond the keyboard.