Interview
Duolingo Co-Founder, Severin Hacker: How AI Impacts the Future of Work and Education
- Duolingo anticipates transitioning to fully personalized, on-the-fly curriculum generation within two to three years, supported by a projected 10x to 100x reduction in support costs that could enable human-grade assistance for the entire free user base despite "Cherven's paradox" of rising demand.
- The company expects real-time AI model costs to decline, eventually allowing advanced conversational features like "AI friends" to become available to all users, while the mission of universal education remains contingent on AI tutors reaching human-tutor quality over a decade-long timeline.
- Content production efficiency is projected to shift from a 12-year cycle for 100 courses to generating 148 courses annually, with future integration of the Duolingo Math app into the main super app to unify the user experience and remove separate download frictions.
- The organization predicts a move away from its current flat structure toward a more hierarchical model with distinct management paths as the company scales, while maintaining a strategy of hiring fresh graduates to better adapt to AI tools and understand the Gen Z demographic.
- Duolingo plans to leverage its "motivation engine" and gamification, specifically loss aversion mechanics via daily streaks and potential future financial stakes, to defend against competitors like OpenAI who may enter the application layer with simplified chat interfaces.
- The market for language learning is forecasted to remain winner-take-all, where high retention metrics are essential to sustain growth and justify paid acquisition strategies, with the Duolingo English Test expected to become the standard objective metric for language proficiency.
- A long-term vision includes the convergence of instruction, credentials, and social connection, though the company acknowledges current gaps in the social domain and aims to solve the challenge of connecting users effectively to disrupt traditional higher education.
- The speaker expects that the "green machine" of experimentation, involving thousands of A/B tests, will continue to drive retention and growth, balancing small daily tests with larger strategic bets while relying on product details rather than a single viral feature for differentiation.
- Duolingo does not plan to open a Silicon Valley office to avoid the internal talent funneling risk posed by local competitors, despite the belief that European founders face a significantly higher probability of success if they relocate to the US ecosystem.
- Regulatory risks from the EU AI Act are anticipated to act as a barrier to innovation, potentially driving founders out of Europe and stifling the startup ecosystem, though the speaker predicts that human-verified content will gain a market premium as a cultural counterweight to AI.
- Future trends suggest a decrease in entry-level barriers for software engineering, allowing non-specialists to build apps, while the "product engineer" role combining product, design, and skills may reduce the need for specialized silos.
- The value of human verification in art and communication is expected to rise, creating a market premium for human-created content, while the "homework cheating" business model faces disruption from AI compared to Duolingo's focus on genuine language acquisition.
- Public market valuation is predicted to increasingly favor predictability and forecasting ability, which may create tension for mature companies if AI disrupts long-term growth trajectory modeling, though private market liquidity may allow consumer brands to delay IPOs.
- Venture capital signaling from top-tier firms remains a decisive factor in fundraising success, and the Duolingo brand is currently viewed as underpriced relative to its global reach, operating efficiently with a marketing team of approximately 40 people.
- The industry outlook includes a future where AGI may occur within 10 to 20 years, potentially leading to job displacement and Universal Basic Income, while schools will likely retain a human element for guidance and coaching despite AI's instructional capabilities.