Duolingo scraps AI performance reviews after employees push back on flawed metrics

Duolingo stops evaluating workers based on how much AI they use

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. Facepalm: In a classic example of a CEO admitting he got too carried away on the AI hype train, Duolingo boss Luis von Ahn has confirmed the company will no longer assess employees' performance based on how much they use AI for work tasks. In April 2025, von Ahn announced plans for Duolingo to become yet another "AI-first" company, meaning more of the technology being integrated into the platform and the eventual elimination of contract workers. This wasn't a first for von Ahn: in January 2024, 10% of Duolingo's contract workers were laid off due to AI adoption. Another element of this new initiative was to evaluate employees based on their AI use in performance reviews. Duolingo certainly wasn't the first company to go down this path: Meta, Google, Microsoft, Shopify, and others do the same. But as most people know, shoehorning AI into every aspect of a business can be a bad idea. Duolingo users' reaction to the AI-first announcement was predictably negative. Duolingo even tried to own its decision using a remarkably ill-conceived Instagram post. A number of users said they would be deleting the app because of the changes. Like many CEOs who seem oblivious to the public's views on AI, von Ahn said he "did not expect the blowback." Just a month after his AI embrace, von Ahn backtracked, saying human workers were still needed and he did not see the technology replacing Duolingo employees. In a recent episode of the Silicon Valley Girl podcast, von Ahn talked about the feedback employees gave regarding AI being tied to performance reviews, with some asking if the company wanted them to use AI for AI's sake. It seems a few weren't sure what was actually being measured, or whether using AI was more important than doing a good job. "At the end, we backtracked, and we said, 'No. Look, the most important thing in your performance is that you are doing whatever your job is as well as possible. A lot of times AI can help you with that. But if it can't, I'm not going to force you to do that,'" von Ahn said. "It felt like rather than being held accountable for the actual outcome, we're trying to just push something that in some cases did not fit." Other companies are also discovering that going all-in on AI can backfire. Klarna is one of the most notable examples - having previously been the most vocal AI advocate, the company started hiring humans again last year after it found AI chatbots offered a "lower quality" output. More recently, Take-Two laid off a portion of its AI team, including its head of artificial intelligence, just two months after publicly saying it would be "actively embracing generative AI."

'We're trying to push something that in some cases did not fit' -- Duolingo's CEO changes course on AI at work

Luis von Ahn admits to trouble translating AI usage into real results * Duolingo CEO says company has scrapped plans to factor AI usage into employee performance reviews * CEO Luis von Ahn admitted to change after significant internal confusion and pushback * Employees worried that the policy risked prioritizing tool use over actual work outcomes Duolingo's CEO promises the company will no longer move toward centering performance evaluations on how much employees use AI for work. Luis von Ahn has repeated a half-embarrassed mea culpa over his initial very vocal enthusiasm in pushing Duolingo toward an AI-first future. AI hasn't left the office entirely, but efforts to measure how much employees use the technology are being scaled back to a more pragmatic level. "At the end, we backtracked, and we said, 'No. Look, the most important thing in your performance is that you are doing whatever your job is as well as possible. A lot of times, AI can help you with that. But if it can't, I'm not going to force you to do that,'" von Ahn said in a new episode of the Silicon Valley Girl podcast. "It felt like rather than being held accountable for the actual outcome, we're trying to just push something that in some cases did not fit." Policy outpaces practice The original idea didn't seem too outlandish at first blush. If AI is going to transform work, employees should use it, and companies should track that usage to ensure they keep up. Duolingo leaned into that logic with an internal push to become "AI-first." The company tested making AI usage a factor in performance reviews. But those same employees began to question what exactly was being measured. Was demonstrating that they were using AI wherever possible, the same as actually improving their work? Tools that were meant to help begin to feel like requirements. The result, according to von Ahn, was a mismatch between what the company intended and what employees experienced. AI reset Duolingo's retreat comes as many companies are grappling with similar questions. Across the tech industry, there has been a rush to integrate AI into everyday work, often accompanied by internal targets or expectations around adoption. What remains unsettled is how to measure success. Counting usage is easy. Measuring impact is harder. AI is still part of the company's strategy, but it is no longer a proxy for performance. Now, von Ahn said, the emphasis is back where most employees expected it to be in the first place. Work is evaluated on quality and results, not on whether a particular tool was used along the way. AI may be powerful, but it is not universally applicable. There are moments when it improves output and moments when it does not. Forcing it into every corner of a workflow risks flattening that distinction. Von Ahn has long argued that AI should enhance human effort rather than replace it, a point he has repeated in various interviews as the company faced backlash over its earlier messaging. His latest comments reinforce that position, even as they acknowledge that the path to integrating AI is not as straightforward as early enthusiasm suggested. The adjustment won't resolve the bigger concerns around how AI is changing work. Still, any shift toward something more sustainable and human-friendly is a good sign. Even if it's just switching from asking employees to prove they are using the latest tools to making sure they are leveraging them when appropriate. Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button! And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

Duolingo was evaluating its workers' AI use. Workers pushed back.

Subscribe to the Modern CEO newsletter.Refreshed leadership advice from CEO Stephanie Mehta It's not the first time the company has backtracked on its AI policies. Last April, von Ahn announced that Duolingo would become an "AI-first" company, even urging teams against hiring more employees unless the team could not automate their work. With the use of AI tools, Duolingo was able to introduce 148 new language courses. But a week and some major social media backlash later, von Ahn backtracked on those comments, and wrote in a LinkedIn post that he doesn't view AI as a replacement for Duolingo employees. In the recent podcast episode, von Ahn seemed to maintain that belief. "Despite what the internet may think, we've never done a layoff here," he said. "The way I see it, a single employee is just way more productive now than they used to be." While von Ahn said that Duolingo staff have become better at leveraging AI, there are some cases in which AI does the opposite of boosting productivity. For example, AI tools are not all that great at generating a large number of narratives, often coming up with things that don't make any sense, he said. Additionally, von Ahn said that "it's not yet the case" that AI can code better than humans. In many cases, the tech doesn't always work well when it comes to programming and when mistakes are inevitably made, they take longer to identify and debug.

Duolingo's AI U-turn Is a Warning for Other Companies

If you work at many big companies these days, finding ways to use AI to do your job better isn't a suggestion. It's a requirement. As The Wall Street Journal recently reported, "From small startups to giants including Amazon.com, Alphabet, Google, and Meta Platforms, tech companies are measuring [AI use] with an eye on productivity gains and in certain cases factoring it into performance reviews." Given the industry's mad dash to realize the potential productivity gains of AI and keep ahead of the competition, leaders' desperation to have employees embrace AI makes sense. But is tracking and scoring AI usage in performance reviews the best way to go about it? The experience of learning app Duolingo, as well as some fascinating recent research, suggests companies should think carefully about how they evaluate employees' AI. The potential for unpleasant and unintended consequences is high.