In a preregistered online experiment, we assigned occupation-specific, incentivized writing tasks to 453 college-educated professionals and randomly exposed half of them to ChatGPT. Our results show that ChatGPT substantially raised productivity: The average time taken decreased by 40% and output quality rose by 18%. Participants assigned to use ChatGPT were more productive, efficient, and enjoyed the tasks more.

Access to AI assistance increases worker productivity, as measured by issues resolved per hour, by 15% on average, with substantial heterogeneity across workers. We find that access to AI assistance increases the productivity of agents by 15%, as measured by the number of customer issues they are able to resolve per hour. In column (1), we show that, controlling for time and location fixed effects, access to AI recommendations increases RPH by 0.47 chats, up 23.9% from their pretreatment mean of 1.97.

Our findings indicate that LLMs can significantly boost productivity among programmers. Productivity (measured by the number of lines of code produced) increased by 55% for the group using the LLM. Approximately one third of this increase was directly attributable to code generated by the LLM. Our findings indicate that the use of gen AI increased code output by more than 50%. However, productivity gains are statistically significant only among entry-level or junior staff, while the impact on more senior employees is less pronounced.

Artificial Intelligence (AI) may be poised to raise productivity across various domains, including writing (Noy and Zhang 2023), programming (Peng et al. 2023), and research and development (Toner-Rodgers 2024; Korinek 2023).[9] This note reviews the extant surveys on AI adoption at both the employee and firm levels.[9] Surveys of workers show between 20 and 40 percent of workers using AI in the workplace, with much higher rates in some occupations like computer programming.[9] Overall, these findings suggest that regardless of measurement differences in the levels, adoption is rising very rapidly both at the individual and firm level.[9]

Generative AI has the potential to significantly improve firms' productivity by improving their workforce efficiency, notably enhancing workers' short-term productivity for specific tasks. Evidence from randomized controlled trials shows that generative AI tools can increase task productivity by between about 15% and 40%, with larger gains often observed for less-experienced workers and for certain types of knowledge work.

Brynjolfsson, Li, and Raymond (2023) found that access to an AI-based conversational assistant increases the productivity of customer support agents (measured as issues resolved per hour) by 14% on average. In a laboratory experiment on consultants employed by Boston Consulting Group, Dell’Acqua et al. (2023) found that productivity on 18 tasks designed to mimic the day-to-day work at a consulting company increased by 12% to 25%. Peng et al. (2023) hired 95 professional programmers to develop an HTTP server in JavaScript. The group that used GitHub Copilot completed the task 55.8% faster, yet there was no effect on whether the task was completed.

Based on studies of real-world generative AI applications, we assume labor cost savings of roughly 25 percent on average from adopting current AI tools.[6] Table 2 lists several studies of AI adoption and summarizes their results.[6] These studies find gains ranging from around 10 to 55 percent, with an average of around 25 percent.[6] We estimate that AI will increase productivity and GDP by 1.5% by 2035, nearly 3% by 2055, and 3.7% by 2075.[6]

Customer support agents using an AI tool to guide their conversations saw a nearly 14 percent increase in productivity, with 35 percent improvements for the lowest skilled and least experienced workers, and zero or small negative effects on the most experienced/most able workers, Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond report in Generative AI at Work (NBER Working Paper 31161). The researchers find that customer support agents utilizing the AI tool increased the number of customer issues resolved per hour by 13.8 percent.

Artificial intelligence tools like chatbots helped boost worker productivity at one tech company by 14%, according to new research from Stanford and MIT that was first reported by Bloomberg.[1] Researchers measured the productivity of more than 5,000 customer support agents, based primarily in the Philippines, at a Fortune 500 enterprise software firm over the course of a year.[1] Improvement was even greater for “novice and low-skilled workers” who got their work done 35% faster.[1]

Generative AI could enable labor productivity growth of 0.1 to 0.6 percent annually through 2040, depending on the rate of technology adoption and redeployment of worker time into other activities. This implies that, at the level of the whole economy, generative AI does not automatically translate into double‑digit yearly productivity gains, but instead into fractions of a percentage point added to growth each year.

This report presents the most recent findings of Microsoft’s research initiative on AI and Productivity, which seeks to measure and understand the productivity gains associated with LLM-powered productivity tools like Microsoft Copilot. One of these is, to our knowledge, the largest, randomized controlled trial of the introduction of generative AI into organizations. Overall, the research suggests that generative AI is already aiding workers in becoming more productive in their day-to-day jobs in significant ways, though the influence of generative AI is subject to variation by role, function, and organization.

Generative AI can improve a highly skilled worker's performance by nearly 40% compared with workers who don't use it.[4] The GPT-only participants saw a 38% increase in performance compared with the control condition (no access to AI).[4] The performance of those who were provided with both GPT and an overview was boosted even higher, with a 42.5% increase in performance compared with the control condition.[4] But when AI is used outside that boundary to complete a task, worker performance drops by an average of 19 percentage points.[4]

On average, across the three studies, generative AI tools increased business users’ throughput by 66% when performing realistic tasks. Study 1: Support agents who used AI could handle 13.8% more customer inquiries per hour. Study 2: Business professionals who used AI could write 59% more business documents per hour. Study 3: Programmers who used AI could code 126% more projects per week. Thus, my early estimate is that deploying generative AI across all business users can potentially increase productivity by around 66%.

According to a 2023 MIT study, generative AI tools have increased writing speed by 40% (Harvard Business Review).[7] Teams using AI report 77% faster task completion, 70% fewer distractions, and a 45% boost in productivity (Hubstaff).[7] According to 2025 data, teams using AI report 77% faster task completion, 70% fewer distractions, and a 45% boost in overall productivity.[7] Additionally, MIT studies show that AI tools like ChatGPT and Copilot have increased writing speed by 40%, demonstrating significant measurable improvements across various work tasks.[7]

The most compelling individual productivity data comes from controlled experiments. A six-month field study revealed that employees using AI tools spent 25% less time on email management and administrative tasks. The study design included randomized assignment and controlled for factors like role, seniority, and email volume. The 25% reduction in email processing time represents one of the most significant and measurable productivity gains documented in 2025.

Its adoption in the workplace has surged as organizations leverage GenAI to boost productivity, streamline decision-making, and foster innovation.[3] This study empirically examines the link between innovative job performance and GenAI tool usage.[3] Analyzing survey data from 366 employees nationwide revealed that both cognitive and social uses of GenAI tools significantly enhance innovative performance.[3] Enhanced knowledge transfer and resource acquisition, in turn, improve job satisfaction, which is pivotal in driving innovative performance.[3]

We conduct a randomized controlled trial (RCT) to understand how early-2025 AI tools affect the productivity of experienced open-source developers. Unexpectedly, our findings reveal that the use of AI tools actually slows developers down, with tasks taking 19% longer compared to when they work without AI assistance. When AI tools are allowed, developers primarily use Cursor Pro, a popular code editor, and Claude 3.5/3.7 Sonnet. Surprisingly, we find that allowing AI actually increases completion time by 19%—AI tooling slowed developers down.

METR previously published a paper which found the use of AI tools caused a 20% slowdown in completing tasks among experienced open-source developers, using data from February to June 2025. Our early 2025 study found the use of AI causes tasks to take 19% longer, with a confidence interval between +2% and +39%. For the subset of the original developers who participated in the later study, we now estimate a speedup of -18% with a confidence interval between -38% and +9%. Among newly-recruited developers the estimated speedup is -4%, with a confidence interval between -15% and +9%.

Macroeconomic projections differ: an IMF working paper pegs about 1.1% cumulative productivity gains over five years under realistic adoption, while McKinsey envisions a far larger upside tied to rapid enterprise deployment.[8] Multiple studies over the years on previous technologies suggest that at 20% penetration you start to see early efficiency gains, but macroeconomic impact takes a bit longer.[8] At about 50% penetration, complementary innovations foment widespread productivity gains.[8]

Bottom line is that developers thought that AI was making them 20% more productive, when it was actually making them 19% less productive. Results are based on 16 experienced developers doing paid work in existing, high quality, open source code bases, in a randomized controlled trial of the productivity impact of AI tools in software engineering for existing code bases. To be clear, there are unambiguous cases where AI tools are providing real boosts to productivity, most notably in rapid prototyping and repeated programming against common APIs.

McKinsey’s analysis reveals that generative AI could automate work activities, absorbing 60–70% of employee time, compared to the previous estimate of 50% for traditional automation technologies. The marketing function demonstrates notable potential, with productivity increases valued between 5–15% of total marketing spending globally. Sales productivity could increase by approximately 3–5% of current global sales expenditures, while retail applications show potential productivity increases of 1.2–2.0% of annual revenues.

Generative artificial intelligence is poised to unleash a powerful wave of productivity growth that will likely affect all industries and could add as much as $4.4 trillion annually to the global economy across the 63 use cases we studied. Research and experiments have shown increases in the productivity of software engineers by 20, 30, 40, 50, maybe even higher percentage points when using these tools, indicating that in specific occupations productivity effects can be very large.

Employees across industries continue to adopt AI tools at a rapid rate, yet the technology's impact on productivity and efficiency is uneven and muddled, according to a new study.[10] 77% of employees report AI has increased workloads and hampered productivity, study shows.[10] Companies mandate generative AI use among employees amid concerns about its actual impact on work quality and efficiency.[10]

The findings indicate that while AI does enhance developer productivity, the extent of this improvement is not consistent across the board. The reported increase in productivity averages between 15-20%, but its effectiveness varies significantly based on the nature of the task, the existing codebase, and the programming language used. The study developed a model to evaluate source code changes based on quality and maintainability, revealing that AI does increase developer productivity, though with nuanced impacts on code quality.

In an eight-month investigation into the impact of generative AI on work practices at a technology firm in the U.S. with approximately 200 staff members, we discovered that employees increased their work speed, expanded their range of responsibilities, and often extended their working hours voluntarily. Employees independently increased their output because AI made it feel achievable, accessible, and, in many instances, intrinsically fulfilling. The study also noted risks of burnout alongside large productivity gains sometimes described as "10x" in qualitative terms.

Across multiple controlled experiments and early field studies, generative AI tools such as large language models, coding assistants, and conversational agents often show double‑digit percentage improvements in task throughput or speed, typically in the range of about 10–60% depending on task type and user experience level. However, effects are heterogeneous: entry‑level or less‑skilled workers tend to see much larger gains, while highly experienced workers sometimes see small gains, no effect, or even slowdowns, especially when tools are new or workflows are still adapting.

According to McKinsey’s latest findings, more than 80% of businesses that have embraced gen AI have yet to realize meaningful improvements in productivity or ROI. The post argues that many companies are stuck using copilots, not AI agents, and claims that some firms are already seeing 90% gains in response time and 15% boosts in code quality from deploying AI agents, although these higher figures reflect selected cases rather than the broad average.

Our latest research estimates that generative AI could add the equivalent of $2.6 trillion to $4.4 trillion annually across the 63 use cases we analyzed — by comparison, the United Kingdom’s entire GDP in 2021 was $3.1 trillion. This would increase the impact of all artificial intelligence by 15 to 40 percent. This estimate would roughly double if we include the impact of embedding generative AI into software that is currently used for other tasks beyond those use cases.

AI agents could boost productivity by 3–5% annually and lift growth by 10% or more, according to a McKinsey discussion post highlighting their estimates of up to $4.4 trillion in productivity gains. The figures cited refer to potential future annual productivity growth and economic output, not a guaranteed minimum 10% productivity increase for every company that adopts generative AI tools.