Why the tech workforce is quietly splitting in two | Annual AI sentiment survey (Noam Segal)
In a Nutshell
The tech workforce is splitting into two roughly equal groups: one feeling amplified and energized by AI, the other feeling destabilized or diminished, with AI identity impact outweighing every other job factor including manager quality. Burnout rose sharply to 55% while optimism fell to 49%, driven by scope creep—doing more for the same pay—rather than job loss fears, which ranked near the bottom of concerns. Designers, researchers, and data analysts are hit hardest with negative NPS scores and highest anxiety, while founders and small-company workers remain most positive; the core advice is to focus AI use narrowly, watch for unsustainable productivity expectations, and prioritize manager relationships.
These notes were generated by AI and may contain inaccuracies.
Burnout at work is a major problem and AI could add to this already overwhelmed employee problem. The honeymoon period with AI is over. When people are asked what they are afraid of, losing their job to AI is actually second to last. What rose to the top is the expectation to do more for the same pay. Burnout is increasing significantly. Optimism is declining. Half of the people in tech are feeling incredible and the other half report their brain is rotting and their work feels worse. Elon had an interesting way of describing it: we are definitely living in a simulation when we are about to start building data centers in space and AI is going to be as smart as human.
Noam Segal worked with Lenny at Airbnb for many years where he was Lenny's research partner. Since then, he went on to be a research leader at Intercom, Twitter, Wealthfront, Meta, Zapier, and Figma. Over the past couple of years, they partnered on multiple research projects. The most ambitious of which has been a tech worker sentiment survey, which they plan to run every year. This is the second time they ran it. This is the largest survey of its kind looking at how people in tech feel about their jobs, AI, layoffs, burnout, and the future of their careers.
This is the second time running this survey, which is unique in the tech industry. Last year when they ran the similar tech sentiment survey, the title was "burnt out but optimistic." They were seeing fairly high levels of burnout but also fairly good levels of optimism about where things are going. They followed up on the burnout point and created the ARMOR framework. This year they surveyed about 6,000 people from product, engineering, design, research, marketing, and pretty much any role in tech.
Elena Vera posted about the AI confidence theater and how in conversations these days in tech everything is dead: engineering is dead, design is dead, SaaS is dead. Obviously that is not true. The results provide the best picture of how people are really feeling about AI. If you work in tech, you probably answer some sort of biannual Culture Amp style survey about how you are feeling, but the questions do not really dive any deeper than the effectiveness of your manager or leadership at the company you work at.
The big picture takeaway is that the workforce is bifurcated right now. AI has an outlier level outsized impact on how people feel about their work more so than any other characteristic of their job, whether it is their role, the company they work for, the company size, their level, etc. Half of the people in tech are feeling incredible, energized, amplified, excited about the technology and the future of their role. The other half feel like the future is unclear. They feel destabilized and diminished.
Only 3% of people reported that AI has not shifted their identity, so clearly AI has changed how people view themselves. 50% of people in tech are feeling amplified. They feel like they can do more and do better. The other half divide into three distinct groups. The first group is people who feel like their role is being redefined. That group does not actually have that much clarity or specific positive or negative emotions around what is happening. They just do not really understand it. It is a very mixed bag type of emotion. That is 27% of the sample.
14% of people feel destabilized. The ground is shaking beneath them. They have very little clarity on what is going on. They have high anxiety levels and report having very negative and pessimistic views of what is to come. 5% of people feel diminished. They feel like the great powers of AI have taken something away from them and something that obviously is never going to come back. AI and models are the worst today that they will ever be. They are only getting better. So people who feel diminished now will only feel more diminished in the future probably.
These buckets correlate super strongly with basically every other measure of how people are doing. In tech, people are obsessed with the notion of statistical significance. But the thing about statistical significance is that if you have a large enough sample you can reach that type of significance regardless of how much the effect matters practically. A better way to look at things is using an effect size measure like Cohen's D. What effect size tells you is practically how much does this thing matter.
Last year a couple of the largest effects they saw were number one, your manager. Who your manager is and how effective they are matters a ton to how much you are enjoying your job and how optimistic you are and how burnt out you are or not. The second large effect they found was the founder happiness effect. Founders, at least people who are currently founders and in their journey of running a startup, are incredibly happy probably because they have a lot of agency and autonomy relatively to the typical employee. This finding around the impact of AI on your identity and on every single other variable around your job is about three times as large as those other effects. This technology and this era is having an outlier level outsized impact on how people are feeling about work more so than anything else seen.
As you move from the people who feel amplified to the people who feel diminished, career optimism goes down significantly. Burnout or reports of burnout go up significantly. Layoff worry also rises significantly. The most significant finding is that people would not recommend junior people, early career people entering into the tech workforce at this moment given what is happening with AI. These are very linear, clear, and organized effects.
They asked people about how they are feeling and to report specific emotions. People could choose as many emotions as they wanted. On average, people selected five different emotions. A couple of people selected 13 emotions that they are feeling.
The first archetype is the energized tech worker and that is 41% of the sample. These are the people who are saying that product has become fun again. They are exploring things. They feel like they are in the tech amusement park where everything is open to them and they can experiment with new approaches. They are a builder now. They have powers that they have never had before. These people are very energized about the technology and about their careers.
The second group is called the conflicted, the ambivalent middle. These are 35% of the sample. These are people who are having on the one hand the most fun they have ever had as builders, as PMs, as designers, as engineers, whatever their role is. And on the other hand, they are feeling the most uncertainty that they have ever felt in their careers. They just do not understand where this is going and whether they are building the things that will ultimately lead to the end of their careers as they know it.
The third archetype are the disoriented people. The people who feel like their role keeps shifting. One respondent said they are like farmers on the cusp of the industrial revolution and they just do not see a clear path to what is happening. These are people who are feeling very disoriented around what is going on in their career and where things are leading.
Finally, 12% of the sample are categorized as resentful. They are feeling pressured. They are feeling checked out. They are not enjoying their work right now. They are saying things like, "I've been forced to use AI or lose my job. And even when I use AI, I'm still seeing people lose their jobs. I just hate it. I hate having to leverage this technology rather than do my own thing and maybe ignore the technology."
Given the pace of change of the technology, too much focus is being spent on mastering the technology and trying to understand what to do with new releases, new models, new capabilities, and prompt engineering. Not enough time and focus is being spent on relationships at work, on seeing colleagues, on supporting peers in also being able to work with this technology, thrive with it, succeed with it, and get along and collaborate well with people who do not belong in the same bucket when it comes to how the technology is being perceived. Energized people should take some of that energy and spend it on the people around them rather than the all-in focus on whatever the next thing is with AI.
Last year they looked at burnout and optimism. Burnout was at a worrying level enough that they published further research on it and introduced the ARMOR framework around how to deal with burnout and fight burnout. This time around, when it comes to significant burnout that is higher than moderate, they have gone up from 44.7% in 2025 to 54.7% in 2026. So burnout is surging. More than half of the sample are feeling significantly burnt out at this point. At the same time, optimism around the future of roles and careers is falling from 54.8% in 2025 to 48.7% in 2026.
Jeff from Ramp shared that his worst burnout happened when velocity was lowest. When you put effort into things that do not actually move, it does not feel good and that leads to burnout. That maps with certain theories around burnout where stagnation leads to burnout. What is going on now is the opposite. As people ship faster than ever, as they move from shipping a couple of PRs a day to 30 PRs a day or just doing so much more, in some ways that might be fun, but they are burning out more than they did before. They are not working any less hard. They are just taking on more stuff: more prototypes, more PRDs, more PRs, more campaigns, more agents, more ads, and it is leading to more burnout.
Enjoyment of work is actually still very high. It is burnout, but also people are still at the same levels as last year really enjoying their work. What was heard about enjoyment more than anything else is that people are able to bring out certain aspects of their identity and of their passions that they were not able to before. If you are currently a PM by role, but you have a designer within you, or you are a marketer, but really inside you have that engineer that has been really wanting to come out, now you can. Various tools that are available enable you to bring out those sides of your personality and those professional aspirations. People are really enjoying escaping their classic swim lanes and a certain percentage of roles are transforming into something else.
The second thing people are saying that they are finding enjoyable is that they are able to manifest things that they have wanted to build that seemed impossible not very long ago, whether it is in their main job or things that they are doing on the side. It is still for many people in tech an exciting time and an enjoyable time because they can do things that they never thought they would be able to do because they are outside of their role, because they are outside of what felt technically feasible, or for other reasons.
72% of the sample are worried to a certain extent about being laid off and 41.2% are at least moderately worried about being laid off. It is clear that people are feeling that they might be cutting the branch that they are sitting on when it comes to using AI. People are reporting that their companies are going all in on agents, all in on these technologies, and it does feel like they may be pushed out at some point. So the worry level is high.
There is a ton of ambivalence right now in the community and a ton of clashes between emotions: feeling energized and enjoying the job and at the same time not being very optimistic about it and feeling tired from it and feeling worried that all of that enjoyment is going to end one day when the company decides that the person is no longer needed.
There is something addictive about this technology. Many people have never had as much fun in their career as when using these technologies. People have been using it for too many hours a day playing about and building. Many people, even those who are not as energized about the technology, still feel like they are in this fascinating technological playground. It is very hard to disengage from that technology and touch grass, read a book, or whatever you love to do in your spare time. People are enjoying using this technology because it has opened up avenues that they have never had. But on the other hand, it is tiring. It is tiring to be constantly learning this technology and constantly trying to apply it in different contexts and constantly trying to take on more than what your traditional role dictated because there are not any traditional roles anymore. Everyone is just builders these days whether that is an official thing within a company or not.
This naturally leads to these clashing emotions: these exciting feelings around what is possible and how fun it is to use AI. And on the other hand, the realization that the reason this technology is so exciting is because it is so powerful. And because it is so powerful, it might be coming for us and to take our jobs.
Noam Segal built a website called NPSisworst.com dedicated to fighting the use of Net Promoter Score. He also owns NPSthebest.com and forwards it to NPSisworst.com.
The survey asked senior professionals whether they would recommend their role or the tech industry to a friend or relative. The NPS scale runs from 0 to 10, with NPS overall calculated from -100 to +100, where zero represents neutrality.
No role in tech showed promoters. Founders, the happiest group, would not recommend their roles. Sales, go-to-market, PMs, operations, engineering, designers, and researchers all showed negative NPS scores. Designers and researchers were the least likely to recommend their roles.
As seniority decreases, recommendation likelihood decreases. Execs, VPs, and directors are more likely to recommend their roles than individual contributors.
One explanation is that VPs benefit more from AI because they receive processed information and knowledge streams. ICs are building micro-SaaS products with duplicative work, unclear visibility across teams, and difficulty maintaining products. ICs describe the feeling as "full gas on neutral" - pushing the pedal to the metal but going nowhere.
A parent noted they previously pushed computer science but now don't know what roles will exist for their kids in a few years. The current uncertainty is new - these results would not have appeared one or two years ago.
Scott Wu of Cognition described Devon progressing from high school CS student to college intern to junior engineer to senior or staff level engineer. The metaphor suggests technology is climbing the ladder and pulling rungs from under people. Those higher on the ladder feel more stable; those lower feel rungs disappearing beneath them.
Sales and PMs were less negative about recommending their roles than expected. Researchers and designers were most negative. The research community has experienced ongoing concerns about having a seat at the table, finding meaning in work, and democratization of research.
Research and design play an important role in raising the ceiling of what's possible, whereas AI has lowered the bar without raising the ceiling. PM is a generalist role, and the current era favors generalists. Sales and go-to-market people may believe sales cannot be done by AI.
Negative recommendation scores reflect how people feel, not the actual future of roles. This survey captures emotions about working in tech, not objective realities.
When asked if they are happy in their current role, most people said yes. The gap is between current satisfaction and extrapolated future outlook - people are less optimistic about what will happen in one, three, or five years.
In last year's survey, people were bullish about AI improving both productivity and quality. This year, 97.2% of people said AI makes them better at their job, with close to 50% saying AI makes them very much or extremely better.
When people say AI makes them better, they mean they can do more faster, not that the quality of work is better. The concerning finding is the cost to thinking and judgment - people report their "brain is rotting" and their "work feels worse."
Cognitive rot describes accepting AI output without applying judgment or thinking, leading to gradual collapse of mind, involvement, and agency. Simon Willis discussed skill atrophy and the need to consciously resist it.
Models are improving but still require more improvement before work quality increases without human involvement. The recommendation is deliberate practice, deliberate thinking, and deliberate judgment to meet rising bars as technology evolves.
Encountering advanced models reduces the sense of being smart and lowers self-efficacy. Solving problems yourself raises the baseline of self-confidence and self-belief; offloading to AI lowers that baseline.
Productivity gains are real, but the quality of the work and the sharpness of the person producing it are taking a hit.
Losing a job to AI ranked second to last among concerns. The top concern is the expectation to do more for the same pay. The second concern is that the pace is becoming unsustainable - both work velocity and the pace of technological change requiring constant learning.
The speed AI unlocked got plowed straight back into expectations. Every game becomes the new baseline and the people expected to hit it are running out of room to breathe.
The top two emotions are positive: curiosity and excitement. Other emotions include overwhelmed, conflicted, relieved, tired, burnt out, uneasy, anxious, and hopeful.
The industry is in a state of ambivalence with a wide variety of emotions.
Nikil Singal discussed the concept of smiling exhaustion, which captures the current ambivalence in tech. The old form of burnout was defined by completely grim, negative emotions stemming from disengagement and exhaustion. In contrast, smiling exhaustion reflects people feeling reborn on one hand—shipping again, building, and creating incredible things with AI during the most exciting time—while facing no off switch, a brutal tempo, and rules that rewrite themselves daily. The relentless pace forces people to smile through the exhaustion they feel.
Half of people are working very hard, often exhausted, yet also energized and going all in. There is a large group loving this time and doing well. The tendency to think in black-and-white terms—viewing people as either full of hype or doomers, haters, or whatever label—oversimplifies reality. Behind the theatrics, everyone contains both the hype person and the doomer within, differing only in the amounts of each. Emotions like excitement and curiosity alongside overwhelm, relief and tiredness, or resentment can coexist even when overall feelings about the technology are positive. All of these emotions are natural and understandable given the times.
Designers and researchers are the most negative. They lead in feeling destabilized or diminished regarding AI identity shift, report the highest levels of feeling tired, overwhelmed, or anxious, show high worry about losing their jobs to AI, and are the least likely to recommend their role to a friend or colleague.
It is not that these functions are in big trouble, but rather how people in them are feeling right now. The industry needs designers and researchers more than ever. Concepts like taste, craft, quality, beauty, value, judgment, and design matter greatly, as highlighted by leaders such as Katie Dill and Jenny Wen, who leads design for Claude at Anthropic. The community should get in and demonstrate the value of these roles because the industry needs them.
AI is currently bad at design, often producing clearly AI-generated slop that lacks the quality people want. This represents an opportunity for design to become a major value add and differentiator as people grow tired of generic outputs. Some issues with AI performing design may be tractable, but creating incredible, novel, creative experiences remains an intractable or much harder problem for the technology.
Data analysts are the most worried about losing their job to AI, exceeding even designers and researchers. This aligns with incredible capabilities emerging from models for running data analysis. Engineering was expected to show higher concern in some areas due to advancements in coding, but feelings do not correlate one-to-one with objective capabilities.
Founders and people working in smaller companies remain the happiest, consistent with last year's survey. Founders show 71% optimism, enjoy their jobs the most, have the lowest burnout, lowest layoff worry, and the most AI excitement. This is a medium-sized but very significant effect. However, selection bias applies: the data reflects active founders of still-running startups rather than those who shut down. Even founders experience limits to ownership and agency—47% are moderately or at least moderately burnt out, and they are not very likely to recommend being a founder to others.
Optimism decreases, burnout increases, layoff worry rises, and likelihood to recommend the role falls as company size grows. Burnout climbs significantly from 1-10 person startups to 5,000-10,000 person enterprises in a strikingly linear pattern with no sweet spot or reversal at larger scales. Even the lowest burnout levels remain fairly high, and the overall level of worry in tech is concerning.
Managers matter tremendously—who your manager is likely matters more than most other role characteristics. The more effective the manager, the less burnt out people feel and the more they enjoy their job. Extremely effective managers correlate with around 65% higher job enjoyment and dramatically lower burnout. Only about 25% of the sample rate their manager as highly effective, while 36% rate managers as ineffective, showing little change from last year.
The era of the great flattening and founder mode means people have more direct reports than ever. While flatness has arguments in its favor, the importance of managers on well-being raises questions about this trend. Managers are the ones who manage the squeeze of overwork from AI, protecting teams or dictating workload. The biggest lever for increasing retention is improving managers and connecting people with better ones. The worst-rated managers cluster in data analytics and design.
Design managers, research managers, and data and analytics managers are themselves suffering, making it hard not to transmit negative energy to their teams. Manager training remains too rare. Leaders should invest more in managers. When job hunting, carefully consider who the manager will be, as nothing impacts well-being more. Poaching and aggressive offers from well-funded labs make it especially difficult to retain trained managers.
Thousands of open-ended responses describing the state of the tech industry clustered into themes of change, chaos, speed, excitement, flux, hype, lack of stability, potential bubble, crazy opportunities, everything evolving, instability, costs, better and worse, huge, and confusing. The industry feels more chaotic than ever. A memorable quote captured it as being "in the second inning of a massive shift. No one knows how it will end, but all you can do is keep taking at bats."
Sentiment analysis showed an even split: 37% positive words, 37% negative words, and 26% neutral. Half find the moment thrilling while the other half finds it terrifying. This may be the most normal it ever gets.
The current period represents an unprecedented amount of change, comparable to the industrial revolution. The odds of being alive during the moment when building data centers in space and AI reaching human-brain-level intelligence become possible are remarkable.
Having the most advanced AI and best models will not determine organizational success. Underlying everything are people going through the most massive shifts in their lives, feeling excited, exhausted, hopeful, and scared often all at once. People are crafting and building a future they hope still has a place for them. Focus should remain on people driving innovations. Individuals should take care of themselves, touch grass, reflect on what they want in their career, recognize that feeling all emotions is fine, and do what works to get through this period.
People who feel amplified and energized went deep on specific tasks and jobs to be done rather than trying to be generalists who do everything. Leaning too far into being a generalist leads to severe burnout. Watch out for the squeeze of being overworked for the same pay. A burnout test is available to assess one's position on the scale.
After reviewing survey results, employees should reflect on their current position and discuss scope adjustments with their manager. Recalibrate and realign work expectations to ensure positive feelings about the role and continued success. Two distinct sets of recommendations exist: five pieces of advice for employees and five for companies.
Pick a couple of specific areas where AI can be useful and go deep on those rather than attempting to apply AI across everything. Focus on the areas most personally exciting and invest significant effort there instead of spreading attention thin.
Watch for being squeezed where scope increases without corresponding compensation changes. Speak directly with managers about situations where workload expands but pay and recognition remain unchanged.
Manager quality matters more than nearly anything else in determining how people feel about work. Protect that relationship, invest in it, build strong communication channels, and learn to manage up. This investment yields significant returns in overall job satisfaction.
Consider opportunities at smaller companies or starting one's own company, as these paths offer distinct advantages. For those earlier in their careers facing disappearing ladder rungs, mentorship remains highly effective. Seek teams, managers, and individuals willing to invest in development, which proves especially valuable in the current environment.
Leaders should invest in their managers, as this represents some of the highest-value spending possible. Only a quarter of survey respondents rate their manager as highly effective, representing a major red flag and significant issue in tech. Improving burnout, enjoyment, and retention amid competitive offers from AI labs requires strengthening management capabilities.
Address the squeeze where AI raises bars in unsustainable ways. Establish appropriate expectation levels and productivity targets. Prevent the bottom rung of the career ladder from deteriorating by supporting entry-level advancement. Early-career individuals often represent the most AI-native talent, using these technologies more naturally than those accustomed to previous paradigms.
Pay particular attention to design, research, and other roles experiencing more negative sentiment. AI lifts some people while destabilizing others, creating unequal experiences across the workforce.
Burnout surveys provide value by helping people identify when their feelings deviate from normal patterns. Recognizing extreme burnout enables informed decisions such as speaking with a coach, discussing with a manager, or taking time off when possible. A science-based quiz will be linked in show notes.
An emerging AI guilt pattern appears particularly among early-career professionals who perceive leveraging the technology as cheating. This guilt decreases with increased seniority. Product marketing, data, and analytics roles report the highest guilt levels. This connects to the established impostor phenomenon where competent people attribute success externally, now applied to AI rather than other people.
AI represents an incredible technology that will only improve. Leverage it and learn to use it effectively, as it shows no signs of disappearing. The current version represents its worst state, making guilt about adoption unnecessary. On the contrary, universal adoption makes avoiding the technology impractical.
Listeners can participate in future research projects including quick polls, in-depth surveys, and interviews. Participation allows voices and perspectives to shape findings shared back with the community. The goal includes helping people perform better at their jobs while feeling better about their work. Despite challenging findings, the outlook remains optimistic with belief in positive directional progress.
Noam Segal can be reached on LinkedIn or within Lenny's community Slack. The full report will be linked for deeper exploration, with a follow-up planned based on conversation outcomes and community comments.
Keep Lenny's Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
