The Cuts

In early January 2024, Duolingo - the language-learning app with over 100 million monthly active users - quietly cut approximately 10% of its contract workforce. The affected workers were primarily content creators: the translators, writers, and language specialists who developed the exercises, sentences, and lesson content that make up Duolingo's courses.

The company confirmed the cuts to Bloomberg and CNN, explaining that it was shifting to use generative AI tools, including OpenAI's GPT-4, to handle more of the content creation work that contractors had done. Duolingo had announced its partnership with OpenAI in March 2023 and had been integrating AI into its content pipeline for months by the time the contractor reductions happened.

The contractor cuts were not technically layoffs of full-time staff - a distinction Duolingo and CEO Luis von Ahn would lean on heavily in the months that followed. Contract workers in the language industry often work on a project basis, without the employment protections and benefits of full-time roles. By reducing contractor headcount rather than firing employees, Duolingo could shift its content production to AI while maintaining that no one had been "laid off."

Former contractors told the Washington Post a different story. They described being cut as the company increasingly used AI tools to generate the content they had previously been paid to create. For the workers who had been writing exercises, translating content, and reviewing lessons for accuracy, the practical effect was the same regardless of what term Duolingo used: their work was being replaced by a machine, and they no longer had jobs.

The "AI-First" Memo

The contractor cuts might have generated modest attention on their own. What turned the story into a major controversy was a company memo from von Ahn that was posted to LinkedIn in April 2025.

The memo described Duolingo's strategy as "AI-first." It outlined a plan to gradually reduce contractor roles that handled "work that AI can handle" and to increase headcount only when "a team cannot automate more of their work." The language was direct. Duolingo was positioning AI not as a supplement to human workers but as the default, with humans brought in only when AI fell short.

The LinkedIn post drew hundreds of comments. Users worried about what an "AI-first" approach meant for the quality of their language courses. Language professionals criticized the framing as dismissive of the expertise required to create effective educational content. Some users threatened to cancel their subscriptions. The backlash was amplified by broader public anxiety about AI replacing human workers across industries - Duolingo became a specific example of a general fear.

Von Ahn responded by telling the Financial Times that Duolingo had been more transparent about its AI strategy than other companies and had underestimated how people would react. He later admitted in an August 2025 Fortune interview that the memo "did not give enough context" and insisted that full-time employees had never been laid off. The company had about 900 full-time employees, he said, and those roles were safe. The contractor reductions were portrayed as natural evolution, not cost-cutting.

The Quality Question

The core question behind the backlash was whether AI-generated content could match the quality of content created by human language specialists. Duolingo's courses span over 40 languages, from Spanish and French to Hawaiian and Navajo. The quality requirements vary significantly by language and by level. A beginner Spanish exercise has different accuracy requirements than an advanced Mandarin grammar lesson. Languages with fewer speakers and fewer available training data (like Hawaiian) pose particular challenges for AI models trained primarily on English and widely spoken European languages.

User complaints about content quality appeared across Reddit, social media, and app store reviews throughout 2024 and into 2025. Common grievances included: repetitive exercises with the same sentences cycling over and over, grammatical errors in generated content, mispronounced words in audio exercises, nonsensical sentence constructions (particularly in less common languages), and the absence of grammar explanations that human-authored content had previously included.

One Reddit user who had been on Duolingo since the app's early days described the quality decline as "dramatic" and specifically linked it to "firing human staff and leaning fully into AI." Other users reported that their paid subscriptions were delivering worse content than the free version had years earlier.

Whether these quality problems were directly caused by AI-generated content or by other changes to the app's curriculum is difficult to establish from user reports alone. Duolingo did not publicly confirm which specific content was AI-generated versus human-authored. But the timing - quality complaints accelerating after the contractor cuts and the AI-first pivot - fueled the perception that the app was sacrificing educational quality for cost savings.

The Business Logic

From a financial perspective, Duolingo's AI strategy made straightforward sense. Content creation for language courses is labor-intensive. Each new exercise, each translation, each audio prompt requires human work to produce and review. At scale, across dozens of languages and millions of exercises, the cost is substantial. If AI tools can generate even half of that content at acceptable quality, the savings are significant.

Duolingo's stock price reflected this. Despite the backlash, the company's shares rose dramatically in 2024 and 2025, driven largely by investor enthusiasm for the AI-first strategy. By the time von Ahn's clarification hit Fortune, Duolingo's market capitalization was around $16.8 billion. Wall Street loved the idea of a content company that could scale its output without proportionally scaling its workforce.

The tension between investor enthusiasm and user concern was stark. Investors saw AI content generation as a path to higher margins. Users saw it as a threat to the quality of a product they relied on for education. Both perspectives had merit, and Duolingo sat in the middle trying to serve both audiences.

The Contractor Problem

The treatment of contractors was the most ethically charged aspect of the story. Language professionals who had built their careers creating educational content found their work suddenly reproducible by AI. These weren't factory workers being replaced by robots assembling widgets. They were subject-matter experts - people with deep knowledge of grammar, cultural context, pedagogical theory, and the nuances of specific languages.

The contractor model made the transition easier for Duolingo to execute and harder for workers to fight. Contract workers typically have fewer legal protections, no severance obligations, and no union representation. A company can reduce its contractor pool by simply not renewing contracts, without the reputational cost of a formal layoff announcement. Duolingo used this structural advantage, and von Ahn's insistence that "no full-time employees were laid off" relied on the distinction.

For the workers affected, the distinction was meaningless. They had been doing the work. Now AI was doing the work. They didn't have jobs.

The "F-r-A-I-days" Detail

One detail from the Fortune interview that drew particular attention was von Ahn's mention of "F-r-A-I-days" - a weekly Friday initiative where Duolingo teams were encouraged to experiment with AI to find ways to work more efficiently. The name, a play on "Fridays" and "AI," captured the company's enthusiasm for the technology. To supporters, it showed a company seriously investing in learning how to use new tools. To critics, it felt like a celebration of the same technology that had eliminated their colleagues' jobs.

Von Ahn maintained that the AI-first memo had not drawn scrutiny internally - that Duolingo employees understood the context and were not worried about their jobs. The backlash, he said, came from external observers who read the LinkedIn post without that context. Whether that accurately reflects internal sentiment is impossible to verify from outside. But the public reaction was clear enough: a company that built its brand on accessible, friendly education had become a case study in AI-driven workforce reduction, and the explanation that only contractors were affected didn't ease the criticism.