GenAI's Impact on U.S. Productivity

Just out: Quantifying the impact of #genAI on job performance, by Erik Brynjolfsson & team: "Access to AI assistance increases worker productivity, as measured by issues resolved per hour, by 15% on average, with substantial heterogeneity across workers. The effects vary significantly across different agents. Less experienced and lower-skilled workers improve both the speed and quality of their output, while the most experienced and highest-skilled workers see small gains in speed and small declines in quality. We also find evidence that AI assistance facilitates worker learning and improves English fluency, particularly among international agents. While AI systems improve with more training data, we find that the gains from AI adoption are largest for moderately rare problems, where human agents have less baseline experience but the system still has adequate training data. Finally, we provide evidence that AI assistance improves the experience of work along several dimensions: customers are more polite and less likely to ask to speak to a manager." Open access: https://lnkd.in/d4UecpnQ

It seems to have become an accepted fact that Generative AI will replace many office jobs. And today I've written in the Times to ask: what if this isn't the case? We’ve all seen the reports and articles detailing the negative impact of #GenAI on our workforces. As a chief executive, I read these and think about the atmosphere of uncertainty they might be creating, damaging morale and demotivating people.  I am not alone in questioning that negative narrative. KPMG’s latest global CEO Outlook survey of more than 1,300 CEOs shows the leaders of the world’s biggest businesses see GenAI as positive for workers. We found that 71% of UK CEOs see GenAI as an opportunity to try new ways of working, creating a highly skilled and productive workforce without significant job losses. A third even think it will create more jobs. We found similar views across the broader population of CEOs globally. And when I look at how we have been introducing AI and GenAI at KPMG UK, I can see that it is making the work we do even better. All our audits are digital. We use AI today to support our teams, and the ambition is for all our audits to always be delivered using the latest technology, including GenAI, as this will lead to even better quality audits. As a graduate auditor I spent many, many - often frustrating - hours transferring data from a ledger. Today our auditors are saved from this grind by our AI enabled tools, freeing them up to spend more time talking to clients and focusing on the more judgmental areas of the audit. Of course, implementing GenAI doesn’t come without challenges. When I talk to other CEOs the same thorny issues come up time and again: trust, regulation, and concerns about a lack of skills. This is why it needs to be implemented with care. For me, human intelligence combined with artificial intelligence are greater than the sum of their parts. With GenAI we have a genuine opportunity to help solve the UK’s productivity puzzle. There’s a role for the new Government to make sure young people are starting out with the right skills. And there’s a role for businesses of all sizes in partnership with government, both national and local, in helping to achieve this.   The GenAI story is changing every day and this is only the beginning. But the leaders who do get it right can look to a motivated workforce, empowered to do more interesting and productive work. That’s what I want for my people. So it’s time to change the story on GenAI and to see it as the great enabler of our time. #AI #Technology #Skills #CEOoutlook

Recently there has been growing skepticism about whether Generative AI's ROI can ever justify current and projected levels of investment, with Goldman Sachs posing the question "too much spend, too little benefit?" My thoughts: Even in its early stages of development, we are seeing #GenAI deliver significant boosts to employee productivity and customer experience. Amongst ServiceNow early adopters of Now Assist, our GenAI solutions, we have seen: 👨🏽💻 IT agents spend up to 30% less time getting to successful resolutions ⚙️ Developers increasing velocity by as much as 25% ✅ Self-service deflection improving by >80% for IT and HR requests Across ServiceNow itself we have achieved $5M+ annualized cost takeout and additional $4M+ in increased productivity, as a direct result of our investments in Gen AI for our internal usage. That’s $10M in tangible benefit. We have been disciplined in developing GenAI solutions that focuses exclusively on practical use cases to help key personas (e.g. IT or HR service agents, developers) in the context of their work. We often engage customer teams and our own employees who fit these personas to test solutions like Now Assist in their everyday work – leading to the results mentioned above. This differs substantially from more generalist approaches to GenAI model training and tuning. We are also seeing that the most common inhibitors to achieving ROI from GenAI are not related to the technology itself, but to the quality of data within the organization. Where data is well-structured, accessible, and consolidated on a single platform environment - such as what ServiceNow offers - the use of GenAI and broader AI tends to yield higher, more sustainable returns. Excessive exuberance about any new technology warrants a little healthy skepticism but when it meets practicality it generates tangible, measurable value. To avoid "too much spend, too little benefit" with GenAI, our philosophy has been to apply it only where it is the most appropriate technology to solve real problems experienced by real people. And it's working!

🚀 𝐈𝐬 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐀𝐈 𝐛𝐨𝐨𝐬𝐭𝐢𝐧𝐠 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡𝐞𝐫𝐬’ 𝐩𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐯𝐢𝐭𝐲? In our new working paper, we identified researchers whose writing showed indicators of GenAI assistance after the release of ChatGPT in late 2022. We then compared their productivity to that of similar non-users. The results are striking: 🧑🔬 Researchers with indicators of GenAI use published 36% more papers in 2024 than comparable non-users. But does higher output mean lower quality? 🤔 Not in this case. We also find a moderate, statistically significant rise in the average journal impact factor of adopters’ publications. This suggests that GenAI boosts productivity without compromising quality. Perhaps most compelling, GenAI seems to help level the playing field in academia. The strongest productivity gains appear among: 📈 Early-career researchers 🌍 Authors from non-English-speaking countries This points to GenAI’s potential to reduce structural and linguistic barriers, fostering a more inclusive global research environment. Overall, GenAI appears to be evolving from a simple writing assistant into a powerful tool for both efficiency and equity in science.

GenAI is emerging as a new engine of US economic performance 🤖 As Lydia Boussour and I highlight in our latest analysis on AI-powered growth, generative AI is now leaving clear, measurable footprints in the data. 💸 AI-related investment in software, R&D and information-processing equipment surged at an 18% annualized rate in the first half of 2025 — contributing about 1pp to Q2 GDP growth. Since 2020, AI-linked investment is up 48%, while non-AI investment has been broadly flat. 📊 Adoption is accelerating. The share of US firms using AI to produce goods and services has jumped from 3.7% to 10% since late 2023, led by information, professional services and finance. ⚙️ Productivity signals are emerging. Frequent AI users report meaningful time savings, pointing to gradual — but real — efficiency gains. 🔍 As rapid GenAI adoption reshapes industries, investment in capabilities, workforce upskilling and digital infrastructure will be critical for competitiveness. And because traditional metrics like GDP understate AI’s full impact, leaders should focus on the underlying transformation rather than the headline numbers. 👇 Want to learn more via EY-Parthenon

This paper on AI from the Fed is the single best thing to read for a snapshot of: 1) The recent innovations in generative AI 2) How firms are using AI 3) The likely macro implications of AI I can't often can bring myself to fully read long reports, but this one is well worth it. Abstract: With the advent of generative AI (genAI), the potential scope of artificial intelligence has increased dramatically, but the future effect of genAI on productivity remains uncertain. The effect of the technology on the innovation process is a crucial open question. Some inventions, such as the light bulb, temporarily raise productivity growth as adoption spreads, but the effect fades when the market is saturated; that is, the level of output per hour is permanently higher but the growth rate is not. In contrast, two types of technologies stand out as having longer-lived effects on productivity growth. First, there are technologies known as general-purpose technologies (GPTs). GPTs (1) are widely adopted, (2) spur abundant knock-on innovations (new goods and services, process efficiencies, and business reorganization), and (3) show continual improvement, refreshing this innovation cycle; the electric dynamo is an example. Second, there are inventions of methods of invention (IMIs). IMIs increase the efficiency of the research and development process via improvements to observation, analysis, communication, or organization; the compound microscope is an example. We show that genAI has the characteristics of both a GPT and an IMI—an encouraging sign that genAI will raise the level of productivity. Even so, genAI’s contribution to productivity growth will depend on the speed with which that level is attained and, historically, the process for integrating revolutionary technologies into the economy is a protracted one. #AI #Fed #rates #macro #economics

💡 New preliminary but promising research provides what appears to be the first causal evidence that GenAI doesn't just boost productivity—it directly increases firm profits. 💡 A large-scale field set of experiments involving millions of users at a major e-commerce platform found that GenAI enhancements to business workflows increased sales by up to 16.3%, with the largest improvements in customer service applications. Across four workflows with positive effects, researchers calculated an annual incremental value of approximately $5 per consumer. 💡 What makes this study particularly valuable is that it shows GenAI can increase actual sales and revenue, not just help employees work faster. The productivity improvements came through enhanced consumer experience—specifically higher conversion rates—demonstrating that GenAI can create value by reducing marketplace frictions. 💡 The benefits weren't evenly distributed: smaller sellers, less experienced consumers, and tail products derived disproportionately larger gains, suggesting GenAI's potential to bridge capability gaps across marketplace segments. 💡 This research addresses a critical question many executives are asking: does investment in GenAI actually translate to measurable business outcomes? The evidence suggests the answer is already yes. Link to the paper: https://lnkd.in/eFXWcJHg #AI #Productivity 

Everyone’s talking about GenAI. Yet, very few are actually unlocking its value. The problem is where and how we’re using it. Over the past year, I’ve seen countless teams deploy generative AI across ops, product, marketing. The pattern? Almost everyone’s using it. Hardly anyone is seeing meaningful impact. McKinsey says 80% of companies have GenAI in play. The ROI? Underwhelming. Why? Two reasons stand out: We’re applying GenAI horizontally but expecting vertical results. Most tools are general-purpose: writing helpers, chatbots, summarizers. Useful. But they rarely integrate into core workflows like pricing, forecasting, or planning. We’ve built tools, not systems. Real productivity gains come when AI is embedded into how work actually happens, not as a one-off add-on. That means redesigning workflows, ensuring context flows seamlessly, and closing the feedback loop. We’re just at the beginning. Tools will improve fast, but simply “using” AI won’t cut it. What drives impact? Pinpointed use cases and weaving AI deeply into decision-making. The signal is getting clearer: smart AI integration beats broad AI adoption every time.

This ILO-backed research found that GenAI’s real impact depends on context, task complexity, and worker experience. Productivity gains ranged from 20–60% in random controlled trials and 15–30% in field settings, but early-career workers (22–25) in AI-exposed occupations saw employment decline by ~13%. The biggest gains? Novices performing simple tasks. The biggest question - what happens when all that "easy" work disappears? For HR/TA leaders, the AI shift isn’t only about automation - it’s about who gets to learn by doing. The entry-level “training ground” may be eroding, making early-career development more vital than ever. #AI #FutureOfWork

Is GenAI going to decimate the product development world? who knows... Meanwhile, Here are some down to earth patterns for how AI is already helping turn the product flywheel: - Use GenAI code assists to enable fuller-stack engineers and reduce tech debt through creation of safety net test automation coverage, refactoring to patterns, replacing legacy code/languages with modern languages more of your staff is familiar with. - This will enable you to organize smaller product/outcome oriented teams - since a smaller set of fuller-stack engineers can cover more with fewer dependencies - Smaller more autonomous teams enables more streamlined processes and improved focus and time in the zone (even without looking at opportunities to streamline product dev processes themselves by using AI) - GenAI can enable cheaper, faster experimentation / discovery (it compresses the truth curve by reducing the cost of pretotyping style product experimentation techniques) - Smaller, more productive teams can build features faster, and can discover/validate even faster using these cheaper experimentation techniques - This allows for more "shots" on goal for the same innovation capacity. - Which leads to faster product market fit, more likely product/feature fit - As well as reduced toil related to "failure demand" (customers who misunderstand the product, product failures, etc.) - Freeing even more capacity to discover/explore/grow/reduce technical debt / improve the architecture - Which creates even more powerful even smaller teams - Which enables another turn at the simplification/streamlining flywheel... - And another turn at the product organization productivity flywheel... As this flywheel turns faster and faster, the product organization delivers better and better products and outcomes in an increasingly sustainable and resilient manner, with product organization operating systems that become simpler and more streamlined over time, rather than more complex. What are some additional ways you're leveraging GenAI to help turn the Product flywheel? NOTE: The product flywheel isn't new. Techniques like the practices from Extreme Programming have been helping product teams get better and better the more they leverage these practices for decades now. I've been thinking more and more about flywheels recently.... more to come...