--- title: "AI took away the typing, not the thinking: AI burnout, job loss and the changing role of developers | Guardian360" description: "A follow-up to last week's piece on AI-assisted development at Guardian360. The speed is real, but so is the human cost: AI burnout, the quiet fear of building the thing that replaces you, and where a developer's value moves when AI writes the code." url: https://guardian360.net/blog/ai-took-away-the-typing-not-the-thinking/ locale: en source: guardian360.net --- [← All posts](https://guardian360.net/blog/) Opinion # AI took away the typing, not the thinking: AI burnout, job loss and the changing role of developers By Jan Martijn Broekhof · 19 October 2026 [Last week I wrote](https://guardian360.net/blog/we-now-build-in-three-months-what-took-us-three-years/) about how a handful of people at Guardian360 rebuilt our Lighthouse platform in three months, adding more new functionality than we used to deliver in over three years. I meant every word of it. But it was only half the story, and I would not be honest if I left it there. Since publishing that piece, I have had a series of conversations with our own developers and with developers and founders at other SaaS companies. Almost none of them were about speed. They were about fatigue. More AI coding agents running in parallel means more context switching. Keeping track of what each agent is doing, checking its output and deciding what happens next costs an enormous amount of energy. And for the first time, I hear people use the term “AI burnout”, about colleagues and occasionally about themselves. Underneath that runs a quieter concern. Many developers carry a small voice in the back of their mind, asking whether they are helping to build the thing that replaces them. If AI develops everything, what is left for them? Both concerns deserve a serious answer. Mine starts with a simple observation: AI took away the typing, not the thinking. And our brains were never built to oversee forty hours of work in a single day. ## Why does working with AI agents feel more exhausting, not less? The promise was that AI would take work off our hands. In a narrow sense it does, but it replaces one kind of work with another. A developer who used to write code all day now supervises several agents, each producing a day’s worth of output. Every one of those outputs has to be understood, checked, and accepted or rejected. The typing disappeared; the judgement multiplied. Research is starting to put a name to this. In March 2026, researchers from the BCG Henderson Institute published a study in Harvard Business Review on what they call “AI brain fry”: mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity. Of the 1,488 full-time US workers surveyed, 14 percent reported it, and overseeing AI turned out to be the most mentally taxing form of AI use. Those most at risk were people using multiple tools or overseeing multiple AI agents: the early adopters and the most enthusiastic users. Workers with brain fry reported 33 percent more decision fatigue, and 34 percent of them were thinking about leaving their employer, compared with 25 percent of those without it. That group of enthusiasts describes the developers I talk to. They are not reluctant; they are excited. Aruna Ranganathan and Xingqi Maggie Ye of UC Berkeley’s Haas School of Business followed around 200 employees at a US technology company for eight months and reported in Harvard Business Review that AI did not reduce their work, but intensified it. Employees worked at a faster pace, took on a broader range of tasks and worked longer hours, often without being asked, in part because AI felt like a partner that could help them get through their workload. Productivity surged, but so did cognitive fatigue and unsustainable hours. The enthusiasm itself becomes the risk. Nobody tells you to stop when the tool keeps saying yes. ## Is “AI burnout” real, or just change fatigue with a new name? The sceptic’s answer deserves a hearing. Every major shift in how software is built has come with complaints about exhaustion. Researchers at Oregon State University point out that burnout among developers also surfaced during earlier disruptions, from the rise of agile to the COVID-19 pandemic. Perhaps this is simply the next wave, and it will pass once tools and working habits mature. I partly agree. Some of what we see is adjustment. But this wave is different in one important way: it targets attention and working memory directly. The same Oregon State study, presented at the ICSE 2026 software engineering conference, surveyed 442 developers and found that GenAI adoption is linked to burnout through higher workload and organisational pressure. The authors describe how AI shifts work from creation towards curation and oversight, work that is cognitively demanding and requires deep expertise, but is harder to see. One participant compared it to a car engine running permanently at 100 miles per hour when it works best at 60. The wider developer community shows the strain as well. Stack Overflow’s 2026 Developer Survey, with 30,903 responses from 169 countries, found that 45 percent of developers describe themselves as complacent in their current job. Within that group, burnout or tech fatigue was the most frequently cited cause. The survey does not prove that AI is the reason, but it shows that a substantial part of the profession is running low on energy at the very moment the pace is accelerating. The Oregon State study also contains encouraging news. Autonomy and learning resources reduced the effect of AI adoption on burnout. Developers who decide for themselves when and how to use AI suffer less than those who are pushed into it. That is something every employer can act on. ## Are developers helping to automate away their own jobs? This is the uncomfortable question, and I do not want to answer it with reassurance. As an entrepreneur and an employer, I am genuinely worried about it. The early data points in one direction. The Stanford Digital Economy Lab’s August 2026 update of its “Canaries in the Coal Mine?” study, based on ADP payroll data, shows that employment of 22 to 25-year-olds in the occupations most exposed to AI is about 19 percent below where it would be had it kept pace with less-exposed occupations, up from 15 percent a year earlier. The gap arises mainly because employers are hiring fewer young people, not because they are letting existing staff go, and the researchers see no sign yet of economy-wide displacement. Software development is one of the fields they name explicitly. The youngest developers are feeling it first. Europe is not exempt. The European Commission’s Joint Research Centre traced AI exposure across the European labour market and found that the cognitive abilities most exposed are attention, comprehension and logical reasoning. Because these abilities are needed in almost every job, AI exposure has risen exponentially across all occupational categories, with high-skilled work more exposed than elementary work. That is not a software story. It is a story about almost every desk in Europe. To be fair, the damage has not yet materialised here. Laura Lebastard and David Sondermann of the European Central Bank analysed a survey of around 5,000 euro-area firms and concluded in March 2026 that AI is not yet replacing jobs; firms that use AI intensively are even slightly more likely to take on additional staff. They were careful to add that the longer-term impact of AI on employment remains less clear. I read that as: not yet, but no promises. ## What happens when AI leaves the screen? My concern does not stop at software, or even at office work. Until now, AI has mostly lived on screens. That is changing. A new generation of AI, so-called world models, learns how the physical world behaves: what happens to an object when you push it, how a room changes when someone walks through it. Combine that with robots, and AI moves from the keyboard to the factory floor, the warehouse and the care home. This is no longer confined to a distant lab. In March 2026, AMI Labs, the Paris-based start-up co-founded by Turing Award winner Yann LeCun, raised 1.03 billion dollars to build world models, Europe’s largest seed round to date. Its first target customers include manufacturers, carmakers, aerospace companies and the pharmaceutical sector, and LeCun has pointed out that a domestic robot needs exactly this kind of common sense to understand the physical world. Even AMI Labs expects that it could take years before world models move from research to commercial applications. I am not going to put a number on how many jobs this will touch, because nobody can. But I would be lying if I said it does not keep me awake. When I think about the impact of AI on work, I no longer think only about developers. I think about almost everyone. I do not have the answer to that. What I do have is a view on what remains valuable, and that view shapes the choices we make at Guardian360. ## What is left for a developer when AI writes the code? More than many fear, though it differs from what it used to be. When AI produces the code, someone still has to decide what should be built, whether the result is correct and secure, and how it fits into everything else. Someone has to answer for it when it goes wrong. That is judgement, architecture, review and accountability, and none of it can be delegated to the agent being supervised. Within Guardian360, our developers are preparing for that role: coaches and mentors of AI agents rather than producers of code. It is not a lesser role. It is a more demanding one, which is exactly why the burnout risk is real. If we ask people to carry more judgement, we also have to give them room to exercise it. That is why we deliberately limit the number of parallel streams. The BCG researchers reach the same conclusion: they recommend upper limits on how many AI agents one person oversees at the same time, and found that self-reported productivity fell once people used four or more AI tools. Autonomy and time to learn are already part of how we work, and everyone takes time to experiment and innovate. Those are precisely the resources the Oregon State study on developer burnout, presented at ICSE 2026, identifies as a buffer against burnout. ## A tool is only a tool Besides coaching AI agents, there is a second shift underway at Guardian360, and it is the one I find most hopeful. Our developers have always worked closely with our partners, the managed service providers (MSPs) who deliver our platform to their customers. That collaboration is about to become far more important. A tool is only a tool, even when you add AI capabilities to it. Nobody buys a platform for its own sake; they use it to achieve something. That is where we want to mean more for our partners. An AI agent already takes manual work off their hands in onboarding, reporting and connecting APIs, and pentest actions that used to be manual work are increasingly performed automatically. Every one of those tasks is time a partner can spend on their customers instead of on our software. And that time matters, because the partner is the one facing the customer. They hold the relationship of trust, which is one of the things that makes partners so valuable. That trust cannot be handed over to a tool. In the coming months and years, our developers will take on more of that work alongside our partners. A developer’s value will be measured less by the code they ship and more by what partners and their customers achieve with it. I should be honest that this does not suit everyone. Not every developer wants to work this closely with partners and their customers, and that is a legitimate choice. But I believe the developers who move towards the outcome, rather than the output, will find the most solid ground under their feet. ## So, where does your value lie? AI took away the typing, not the thinking. The speed is real, and so is the price our people pay in attention. Pretending otherwise helps nobody, least of all the developers who are tired and quietly worried about their future. If AI writes the code, where does your value lie? And if you lead a team, are you protecting the people who have to answer that question every day? ## Sources - Bedard, J., Kropp, M., Hsu, M., Karaman, O. T., Hawes, J. and Rosen Kellerman, G., “When Using AI Leads to ‘Brain Fry’”, Harvard Business Review (March 2026) - Ranganathan, A. and Ye, X. M., “AI Doesn’t Reduce Work, It Intensifies It”, Harvard Business Review (9 February 2026), [https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it](https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it) - Feng, Z., Afroz, S. and Sarma, A., “From Gains to Strains: Modeling Developer Burnout with GenAI Adoption”, ICSE-SEIS ’26 (April 2026), [https://doi.org/10.1145/3786581.3786934](https://doi.org/10.1145/3786581.3786934) - Stack Overflow, 2026 Developer Survey (6 October 2026), [https://survey.stackoverflow.co/2026](https://survey.stackoverflow.co/2026) - Brynjolfsson, E., Chandar, B. and Chen, R., “Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence”, Stanford Digital Economy Lab (revised 12 August 2026), [https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/](https://digitaleconomy.stanford.edu/publication/canaries-in-the-coal-mine-six-facts-about-the-recent-employment-effects-of-artificial-intelligence/) - Casas, P., Fernández-Macías, E., Martínez-Plumed, F., Gómez, E., González Vázquez, I. and Salotti, S., “Revisiting the occupational impact of AI in the generative AI era”, JRC Working Papers Series on Labour, Education and Technology 2026/02, European Commission (13 March 2026), [https://publications.jrc.ec.europa.eu/repository/handle/JRC145832](https://publications.jrc.ec.europa.eu/repository/handle/JRC145832) - Lebastard, L. and Sondermann, D., “Artificial Intelligence: friend or foe for hiring in Europe today?”, The ECB Blog, European Central Bank (4 March 2026), [https://www.ecb.europa.eu/press/blog/date/2026/html/ecb.blog20260304~d9e34fc95f.en.html](https://www.ecb.europa.eu/press/blog/date/2026/html/ecb.blog20260304~d9e34fc95f.en.html) - “Yann LeCun’s AMI Labs raises $1.03B to build world models”, TechCrunch (9 March 2026), [https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/](https://techcrunch.com/2026/03/09/yann-lecuns-ami-labs-raises-1-03-billion-to-build-world-models/) - “Yann LeCun’s Paris-based AI world model startup raises more than $1BN”, Tech.eu (10 March 2026), [https://tech.eu/2026/03/10/yann-lecun-s-paris-based-ai-world-model-startup-raises-more-than-1bn/](https://tech.eu/2026/03/10/yann-lecun-s-paris-based-ai-world-model-startup-raises-more-than-1bn/)