AI drives workplace productivity up, but employee workload quietly intensifies as breaks disappear

AI Might Make Your Job More Fun

I was sitting at my kitchen island a couple of weeks ago teasing out the implications of disruptive innovation theory for the artificial intelligence industry as my wife and two of our close friends were getting ready to watch the series finale of Andor. I knew I should close the laptop, but every search turned up one more connection and every edit sharpened the argument. I was using AI tools to help with the research, and they had turned the most tedious part of writing into something genuinely fun. So much so that stopping felt like climbing off a roller coaster just before the first big drop. I eventually did watch Andor that day, but it took way more effort to tear myself away from work than it should have. This wasn't (just) a personal failing. Researchers at the University of California, Berkeley's Haas School of Business spent eight months studying what happens when workers at a tech company get access to AI tools. They found that workers moved faster, took on broader tasks, and extended their hours into evenings and early mornings. Nobody told them to, and the company didn't mandate the use of AI. These workers did more because AI made it, in the words of the researchers, "intrinsically rewarding." They described feeling like they had a "partner" that helped them push through their workload. They didn't just do more in less time - they did more in more time, of their own volition. It seemed like every headline tied to the study framed it as a cautionary tale about burnout. Although that's true, it also tells us something surprisingly hopeful about AI and the future of work. To understand why, you have to go back to 1960. That was the year Douglas McGregor published The Human Side of Enterprise, possibly my favorite book on management. He argued that most companies were built on a false model he called Theory X: People are inherently lazy and must be coerced, monitored, and threatened into working. His alternative, Theory Y, proposed something almost radical, which was that people want to contribute. If the conditions are favorable, work can be as natural as play. People don't need to be driven; they need to be unleashed. Managers loved Theory Y...in theory. In practice, most companies still run on Theory X. Gallup has measured worker engagement for two decades. The number has barely moved. Only about a third of American workers report being engaged in their jobs. Engagement programs, culture consultants, office perks, motivational speakers - billions of dollars spent. Needle stuck. And now AI, without trying, has created a Theory Y experiment. Strip away the tedious parts of a job and what happens? The Berkeley researchers found workers filling their breaks, evenings and weekends with tasks they'd been putting off. Not because they had to, but because they wanted to and because the work had become worth doing. You can see this most vividly in software. Developers describe AI-assisted coding in the language of addiction. One programmer wrote that since the arrival of AI tools, his productivity had increased fivefold, but his ability to disconnect had dropped proportionally. Another called it "coding's equivalent of methamphetamine." These are not people being coerced; they are people who can't stop. Here is where it gets ironic. Although AI is creating the conditions for Theory Y from the bottom up, too many managers are using it in service of Theory X. The employee monitoring software market is booming. AI-powered "bossware" now tracks keystrokes, captures screenshots, scores productivity with algorithms, and flags what it considers unusual behavior. The same technology that is making workers love their jobs is being deployed by managers to surveil them. The data on this approach is grim. Large majorities of employees say surveillance doesn't improve their productivity and that it damages trust. According to one analysis, 42% of monitored employees plan to leave within a year, compared with 23% of their unmonitored peers. Companies are spending billions of dollars on tools that actively destroy the engagement that AI could be creating for free. McGregor wouldn't be surprised. Theory X management creates the problem it's meant to combat, he argued. If you treat people as lazy then they will live down to your expectations. Today that self-fulfilling prophecy is turbocharged by neural networks. The lesson of the Berkeley study isn't complicated. AI can boost engagement or kill it - and which one won't be decided by computers. Managers can give their people a tool and watch them grow or point it at them like a camera and watch them update their resumes. But even the optimistic version deserves a caveat. The Berkeley researchers found cognitive fatigue accumulating over the eight months they observed. Sign up for the Bloomberg Opinion bundle Sign up for the Bloomberg Opinion bundle Sign up for the Bloomberg Opinion bundle Get Matt Levine's Money Stuff, John Authers' Points of Return and Jessica Karl's Opinion Today. Get Matt Levine's Money Stuff, John Authers' Points of Return and Jessica Karl's Opinion Today. Get Matt Levine's Money Stuff, John Authers' Points of Return and Jessica Karl's Opinion Today. Bloomberg may send me offers and promotions. Plus Signed UpPlus Sign UpPlus Sign Up By submitting my information, I agree to the Privacy Policy and Terms of Service. A separate study by the AI safety nonprofit METR found that experienced developers were actually 19% slower when using AI tools, despite believing they were 24% faster. Tedium, it turns out, used to serve as a natural circuit breaker. You went home at six o'clock in the evening because the work was unbearable. Now you look up and it's midnight. McGregor's Theory Y always assumed that self-directed workers would direct themselves well. That assumption needs updating for an age when the technology is so engaging that it overrides the body's own signals. For more than six decades, management has assumed that the fundamental problem with work is that people don't want to do enough of it. But the problem was never motivation. It was the work itself. Fix the work, and people won't stop doing it. Which turns out to be its own kind of problem. But it's a much better one than the alternative. And right now, most companies are spending on the wrong one. More From Bloomberg Opinion: * 'AI-Washing' Job Cuts Is Corrosive and Confusing: Gautam Mukunda * Why Bigger Isn't Always Better in AI: Catherine Thorbecke * Anthropic Isn't Exaggerating About an AI Panopticon: Dave Lee Want more Bloomberg Opinion? OPIN <GO>. Or subscribe to our daily newsletter.

AI can double output. Human biology can't | Fortune

In recent weeks, Accenture made headlines for linking senior managers' promotion prospects to their use of internal AI tools. In a market defined by automation and efficiency, employees are expected to integrate AI into their daily workflows. Usage can now shape career trajectory. That policy reflects something larger unfolding across corporate America. Companies are not just using AI to automate tasks. They are using it to raise expectations about how much work humans should produce. This is not inherently misguided. Measurement is essential to discipline and performance. AI tools can reduce friction, eliminate low-value tasks, and clarify goals. Used thoughtfully, they can enhance human capability. The mistake lies elsewhere. The danger emerges when higher measured output is mistaken for sustainable performance. When organizations equate productivity gains with permanent increases in expectation, they effectively borrow against biological reserves. The debt is paid later in disengagement, turnover, and diminished adaptability. AI can double output. Human biology cannot. The logic driving escalation is understandable. If generative tools allow a consultant to analyze twice as much data, why not adjust targets? If coding assistants compress development timelines, why not reset delivery schedules? If dashboards quantify performance in real time, why not calibrate expectations with precision? The problem is that machine acceleration does not automatically expand human capacity. Human performance follows nonlinear curves. Moderate stress sharpens attention. Chronic stress degrades memory, judgment, and emotional regulation. Energy is finite. Recovery capacity is finite. Emotional bandwidth is finite. When AI increases the pace and volume of work, the biological system does not scale in parallel. Technology can compress tasks. It cannot compress recovery. When companies use AI to process twice as much information, attend twice as many meetings, and produce twice as many deliverables, the temptation is to treat that surge as the new baseline. What was once exceptional becomes expected. What was once temporary becomes permanent. Over time, that mismatch produces predictable consequences. Burnout cycles increase. Absenteeism rises. Creative problem-solving narrows as cognitive load accumulates. Discretionary effort declines. The very tools designed to unlock productivity begin to erode the capacities that sustain it. These effects carry measurable economic consequences. Turnover is not a cultural inconvenience. Replacing skilled knowledge workers can cost a significant percentage of annual compensation once recruiting fees, onboarding time, lost productivity, and team disruption are included. If AI-driven expectation resets increase attrition even modestly, the financial gains from higher throughput can be quickly offset by replacement costs and weakened institutional memory. Productivity volatility also affects earnings quality. Workers operating near physiological limits tend to produce short bursts of elevated output followed by fatigue, disengagement, or extended leave. That volatility complicates planning and weakens operational predictability. In knowledge-intensive industries, sustainable value depends less on raw throughput and more on judgment, innovation, and collaborative problem-solving. Those capabilities degrade when biological constraints are ignored. The borrowing-against-biological-reserves dynamic resembles financial leverage. When companies increase debt without strengthening underlying cash flow, they amplify short-term returns but raise long-term fragility. Escalating output expectations without reinforcing recovery, autonomy, and trust creates a similar imbalance. Organizations may post impressive quarterly gains while quietly depleting the human capital that supports future performance. There are also compliance and reputational exposures. As firms collect more behavioral and biometric data through AI systems and wearable technologies, regulators are paying closer attention to privacy and disability protections. A breach involving health or behavioral data can translate quickly into reputational damage and market value erosion. Human capital governance is increasingly part of fiduciary oversight, not a peripheral human resources issue. None of this suggests abandoning metrics. The distinction lies in how they are used. AI should remove friction, not permanently raise the biological ceiling. It should expand strategic capacity, not compress recovery time. Metrics can discipline performance, but they cannot eliminate physiological constraints. Trust plays a decisive role. High-trust environments reduce coordination costs and accelerate execution. When monitoring feels transparent and supportive, adoption tends to follow. When it feels extractive, stress responses increase and intrinsic motivation declines. Surveillance may increase visible output in the short term, but it can quietly raise the long-term cost structure of the organization. Investors are increasingly scrutinizing workforce stability and resilience as drivers of durable performance. Human capital disclosures now sit alongside financial statements in evaluating long-term value creation. A strategy built on doubling output through AI without reinforcing recovery, autonomy, and trust risks creating brittle organizations that fracture under pressure. Boards and executive teams should be asking more rigorous questions as AI adoption accelerates. Are productivity gains coming from friction removal or expectation escalation? Are recovery cycles built into performance systems? Are we strengthening human capital durability or consuming it for near-term gains? Over a three- to five-year horizon, which approach produces more stable returns? The companies most likely to succeed in the AI era will not be those that demand the largest productivity multiples. They will be those that align technological acceleration with biological sustainability. That requires design discipline. It means building recovery cycles into performance systems. It means measuring value over multi-year horizons rather than rewarding quarterly spikes. And it means recognizing that while AI can expand analytical capacity and compress timelines, it cannot rewrite the limits of human physiology. Organizations that ignore that constraint may achieve impressive short-term gains. They may also discover that the true bottleneck in the age of artificial intelligence is not technological capability. It is the biological system expected to keep up with it.

Companies Are Making a Major Mistake With AI Adoption -- and It's Driving Away Their Top Talent

When you give people time back, retention improves, quality increases, innovation accelerates and recruitment gets easier. There's a dangerous assumption spreading through corporate America: If AI saves your team 10 hours a week, you should fill those 10 hours with more work. Double the output. Maximize efficiency. Squeeze every drop of productivity from your newly "augmented" workforce. This thinking isn't just wrong -- it's actively destroying the very advantage AI is supposed to create. As leaders rush to justify their AI investments with measurable productivity gains, they're optimizing for the wrong metric. They're counting tasks completed instead of measuring what actually drives business value: strategic thinking, creative problem-solving and the human judgment that no algorithm can replicate. The companies that win the AI era won't be those that use it to extract more labor from their people. They'll be the ones that use it to extract more humanity from their work. Let's talk about what "2x productivity" actually costs. Replacing a skilled employee costs 50-200% of their annual salary when you factor in recruiting, onboarding, lost institutional knowledge and productivity gaps. Gallup estimates that burnout costs the global economy $322 billion annually in turnover and lost productivity. The Society for Human Resource Management found that 44% of employees cite burnout as a reason for leaving jobs. Now imagine you've implemented AI tools that genuinely save your team 10 hours per week. That's 520 hours annually per employee -- the equivalent of three full months of work. If you immediately reallocate that time to "higher-value tasks" (read: more work), you haven't reduced their workload. You've just raised the baseline expectation. What happens next is predictable: Your best people -- the ones who adopted AI fastest and generated the most savings -- become the ones you lean on hardest. They become the victims of their own efficiency. And within 18 months, they're updating their LinkedIn profiles. The irony is brutal: The AI tools meant to make work sustainable are being weaponized to make it more extractive. The industrial-era equation of productivity -- output divided by input -- made sense when work was repetitive and measurable. Manufacturing widgets. Processing forms. Answering support tickets. But knowledge work doesn't scale linearly. A developer who writes cleaner code isn't just "more productive" -- they're preventing future technical debt. A product manager who thinks deeply about user needs might launch fewer features but create more value. A strategist who has time to synthesize market signals makes better decisions than one churning out rushed analysis. AI's real value isn't helping people do more tasks. It's helping them do better work. Consider what actually differentiates high-performing teams: None of these appear on a productivity dashboard. All of them determine whether companies thrive or plateau. Forward-thinking leaders are adopting a different approach to AI integration: 1. Automate the "chore" work Use AI to eliminate administrative drudgery -- the data entry, meeting summaries, email formatting, calendar coordination and status updates that consume 30-40% of knowledge workers' time. These tasks are necessary but not differentiating. They keep the machine running but don't move it forward. One executive I know implemented AI note-taking and summary tools across her team. The time saved wasn't dramatic -- about three hours per person weekly. But the mental load reduction was significant. People stopped dreading meetings because they knew they wouldn't spend the next hour transcribing and distributing notes. 2. Elevate the "real" work Redirect energy toward what machines can't do: nuanced judgment, empathetic communication, creative problem-solving and strategic synthesis. This is where humans create disproportionate value. A financial services firm used AI to automate its standard client reporting, saving analysts roughly 12 hours weekly. Instead of assigning more clients, they asked analysts to spend that time on deep-dive research and relationship building. Client satisfaction scores increased 23% within six months. Retention improved. The analysts weren't working more -- they were working on what actually mattered. 3. Reclaim time as the reward Here's the controversial part: If someone uses AI to finish their work efficiently, the reward shouldn't be more tasks. The reward should be time -- mental space, reasonable work hours, energy to engage with family and interests outside work. This isn't soft. It's strategic. Sustainable performance requires recovery. Creative thinking requires mental space. Good judgment requires people who aren't perpetually exhausted. Leading organizations are establishing new norms: Measuring outcomes, not hours: If AI helps a team deliver a project in three weeks instead of six, the question isn't "What else can they do in those three weeks?" It's "Did we get the outcome we needed, and is the team positioned for the next challenge?" Protecting boundaries: Some companies are implementing "AI dividend days" -- when teams hit efficiency milestones using automation, they earn flexibility in how they structure their time. Others are explicitly stating that efficiency gains should not increase baseline workload expectations. Rewarding efficiency differently: Traditional performance management penalizes efficiency -- finish your work fast, get more work. Progressive companies are decoupling compensation from time spent and focusing on the impact created. In tight talent markets, this approach isn't altruistic -- it's competitive strategy. The companies attracting top talent aren't those promising unlimited growth opportunities (often code for unlimited work). They're promising meaningful work, reasonable boundaries and the ability to use technology to make life more human, not less. When you give people time back: The question every leader should ask isn't "How do we use AI to get more output?" It's "How do we use AI to make our team's work more sustainable, more meaningful and more human?" We're at an inflection point. The decisions leaders make about AI adoption in the next 24 months will define organizational cultures for the next decade. One path leads to a productivity arms race where AI becomes just another tool for extraction -- squeezing more output until people break. The other leads to a fundamental reimagining of what valuable work looks like and how technology can elevate rather than exhaust the humans using it. The companies choosing the second path will win the war for talent. They'll build cultures where people want to stay. They'll create space for the kind of thinking that drives real innovation. Because if we're not using technology to make our lives more human, we're doing it wrong. We don't need to do more. We need to do better.

AI push speeds up work, but quietly shrinks breaks

With AI's quick help, employees tempted to work through breaks between tasks. Top human resources executives and consultants said early signals of the shift are visible as generative AI tools move from experimentation to everyday work. After months of deeper engagement with artificial intelligence (AI), a SaaS developer realised that the technology that reduced his workload was quietly erasing his breaks and stretching his work hours. The quick help from AI had tempted the employee to use it during short breaks-such as waiting for a file to load, between meetings, or before lunch-letting work continue instead of fully stepping away. None of it felt like extra work. However, over time, the breaks that once alleviated the health impact of prolonged sitting had diminished. It was what researchers describe as indicators of "workload creep", when productivity gains from automation translate not into reduced effort, but into higher targets, tighter timelines and greater cognitive demands. Top human resources executives and consultants said early signals of the shift are visible as generative AI tools move from experimentation to everyday work. At healthcare technology firm Innovaccer, the focus of AI adoption is to shift the cognitive load of work upward rather than simply accelerating tasks, said Satyajit Menon, the company's global head of people experience. Once embedded, the pace of work accelerated rapidly. Innovaccer had to consciously reinforce prioritisation and manager check-ins to ensure that increased speed didn't quietly turn into increased expectations. Menon said the transition initially created friction as some workers resisted AI adoption, while teams faced tighter objectives and key results (OKR) and faster turnaround targets. "Productivity gains often make us a bit greedier, because we want more," he said, cautioning that efficiency gains do not automatically translate into better employee experience.Managers, meanwhile, are emerging as a key pressure point in the AI transition.Amit Khanna, partner at Grant Thornton Bharat, described adoption patterns across organisations as a U-shaped curve, with senior leaders and junior employees using AI more frequently than mid-level managers. For technology and engineering teams, the productivity impact is already measurable. Dhirendra Nath, chief human resources officer at digital business enabler Altimetrik, said that where AI is effectively integrated into engineering workflows, such as faster drafting, coding assistance, test creation, documentation, and triage, it is driving significant efficiency gains. "We're seeing approximately 20-30% productivity improvement. The biggest shift is faster time-to-first-usable output and quicker iteration loops, rather than an immediate reduction in total workload," said Nath. These shifts are also reshaping workforce structures. "AI is materially increasing throughput per employee," said Anurag Malik, partner at EY India.

The AI Story - From Taking Over Jobs to Making People Work Harder

Ever since the AI story caught our collective imagination, it also appears to have generated more chaos than clarity - thanks to the so called AI thought leaders shooting blanks If you aren't confused around the world of AI, you aren't trying hard enough! Old-time tech writers told me that artificial intelligence and machine learning existed for decades. The modern ones said AI got a new lease of life on the back of ChatGPT. Still others spread fears that the frenetic pace of AI development would render many of us jobless. A global CEO even insisted that all promotions would be linked to AI knowledge - but if AI does all jobs, who do they promote? However, our confusion got further confounded last week when a new research noted that AI is actually forcing employees to work harder instead of making their jobs easier. Yes, you read that right. New research from ActivTrak of more than 1.64 lakh employees' digital activity work suggested that "the workday is shrinking but the workload isn't. It says that after examining their digital activity 180 days before and after they started using AI, suggested that AI "intensified" their jobs in almost every category. The time spent on email, messaging, and chat apps more than doubled while use of business software spiked by a whopping 94%. And the irony is that the spike came at the expense of time that staff spent on focussed, uninterrupted work which fell 9% for AI users and stayed same for the rest. However, there is a sweet spot of AI usage that the study has found. Employees spending between 7 to 10% of their total working hours on AI showed the maximum productivity. But, only three per cent of all AI users came within this range. ActivTrack's chief customer officer Gabriela Mauch says, "It's not that AI doesn't create efficiency. It's that the capacity it frees up immediately gets repurposed into doing other work." By the way, this isn't the only study that questions AI's effects on workplace habits. Another study published by Harvard Business Review also said AI isn't reducing work, but intensifying it. Researchers Aruna Ranganathan and Xingqi Maggie Ye spoke of "Workload Creep" which is nothing but staff taking on more tasks than what is sustainable for them to keep doing. This, they said, can create vicious cycle that leads to fatigue, burnout, and lower quality work. In short, AI tools created a vicious cycle: it "accelerated certain tasks, which raised expectations for speed; higher speed made workers more reliant on AI. Increased reliance widened the scope of what workers attempted, and a wider scope further expanded the quantity and density of work." Which brings us to the question of how enterprises plan to expand AI usage in their respective companies, starting with divisional heads pushing the agenda and opening up the data silos that have perennially existed in them. However, what begs a question is when the CEO perks up and tells all staff that their promotions would be linked to their ability to use AI. In a recent episode of "Rapid Response" podcast, Accenture CEO Julie Sweet said AI proficiency is a mandatory part of working at the company and moving up the ranks. "If you want to get promoted, you've got to do the things that we do in order to operate Accenture," Sweet said. What we also know is that last September the company initiated a whopping $865 billion in a 6-month business optimization program that included staff reskilling. Sweet was quite tough in her approach that the company gave a three-year gestation period to staff in order to update themselves with AI tools. Can't blame her, as Accenture began a three-year, $3 billion push back in 2023 to integrate an AI-first approach. One of their targets was to double their AI talent to 80,000 professionals via hiring, acquisitions and training. Does this mean, Accenture's entire 7.7 lakh staffers have to be AI proficient in the new order? If so, what about the job redundancies that use of agentic AI solutions are likely to bring? Maybe, it is just the way Julie Sweet is seeing things. A Gallup research revealed that in Q4 of 2025, only 38% of companies reported integrating AI to improve workplace productivity. Sweet sees it as an extension of the changeovers that happened some decades ago. She believes integrating AI into the workplace is a natural corollary to how computers became word processors at the workplace. When typing classes were replaced by computer skilling, which is now being reskilled to include AI knowledge. Which now brings us to the question of the AI job apocalypse. For some years now, the markets have predicted a major job market disruption and reports of several tech giants preparing for rather large layoffs has also hit the headlines in recent times. However, a recent report released by Anthropic, which believes the worst of it is still some time away. The study, published a week ago by Anthropic, suggests that alarm bells may take some more time ringing for one reason. While "AI is far from reaching its theoretical capability" and "actual coverage remains a fraction of what's feasible," several jobs could be at a higher risk of "AI led displacement" than earlier thought of. Some of the jobs that could be at risk include computer programmers, customer service representatives, data entry operators, medical record specialists, and market research analysts, as these were the "most exposed" to AI's capabilities, the post said. There are others like investment analysts, software quality assurance (QA), and information security analysts, who could also face the axe, once AI fulfils its own destiny. Another factor that the study brought out is that even in these professions, the impact was "more likely to be on older, female, more educated, and higher paid" sections of the staff. As for those at considerably lower risk, these include professions that need a physical presence such as cooks, mechanics, bartenders, attendants etc. Anthropic called them "noble careers" but not paying particularly well. Maybe, there's a story out there for all of us that AI is unravelling slowly. Maybe, there's a story out there for all of us that AI is unravelling slowly. Maybe, there's more than one story. Anthropic itself says that given all the noise around job losses it is tough to gauge how many were actually lost due to AI or for other reasons. One example is how many companies used AI as a reason to downsize the staff they had over-hired in the Covid aftermath. Amidst all these counter-intuitive claims being made by trigger happy experts and so-called business thought leaders, there's one thing that has emerged strongly. That higher education and college degrees would require to be reworked to an extent where every student may have to compulsorily get a hang of AI tools while gaining knowledge of their subject of choice. Which brings us to the really tough question - how many of our academics are actually proficient in the use of AI tools? Come to think of it, how many of us journalists use AI intelligently and diligently to free up some of our time and create better outcomes?

AI may actually push employees to work more, study warns

While AI improves efficiency, the saved time is often filled with additional tasks. AI has always been promoted as a tool that can reduce the daily work pressure of the users. Many technology leaders have also, from time to time, predicted that these smart AI tools will free workers from routine tasks and give them more time for creativity and strategic thinking. However, the new research suggests that the opposite may be happening, as instead of reducing the workload, AI is pushing many employees to work faster and handle more tasks in a day. The findings come from a large analysis of workplace digital activity that tracked how employees used their time before and after adopting AI tools. The results show a clear pattern. While AI helps people work more efficiently, it also encourages them to take on additional work rather than slow down. ActivTrak, a cloud-based workforce analytics and productivity management software company, examined the digital work patterns for about 164,000 employees across 1,111 organisations. Researchers compared workers' activity for 180 days before and after they started using AI tools. They found that work activity increased across nearly every category. The company clocked 443 million hours of recorded work activity. Also read: Meta plans to lay off 20% of staff as AI costs rise: Report The results of the findings show that the time spent on email, messaging and chat applications more than doubled after workers began using AI. The use of business management tools such as human resources and accounting software rose by 94 per cent. At the same time, time spent on focused and uninterrupted work declined. According to the study, the amount of time AI users devoted to deep concentration dropped by 9 per cent, while non-users showed almost no change. "It's not that AI doesn't create efficiency," said Gabriela Mauch, ActivTrak's chief customer officer and head of its productivity lab. "It's that the capacity it frees up immediately gets repurposed into doing other work, and that's where the creep is likely to happen." Also read: Poco M8 price dropped under Rs 17,500: Check deal platform and more Other research points in a similar direction. An ongoing eight-month study of about 200 employees at a technology company found that generative AI tools did not reduce workloads. Instead, employees worked faster, handled a wider range of tasks and often ended up working longer hours. People may take on more tasks because AI makes them feel easier to start, said Aruna Ranganathan, an associate professor at the University of California, Berkeley's Haas School of Business. "AI makes additional tasks feel easy and accessible, creating a sense of momentum," she said. ActivTrak also found that AI use at work is rising quickly. About 80 per cent of employees now use AI tools on the job, compared with 53 per cent two years ago. However, most workers still spend only a small portion of their time using them.