If 2023 was the year of AI hype and 2024 the year of experimentation, 2025 has been the year of infrastructure. By December 2025, Artificial Intelligence in education (EdTech) has ceased to be a novelty; it is now the invisible plumbing that powers global learning. From Duolingo’s content factories to teacher-facing coaching tools, the sector has shifted from “disrupting” education to “sustaining” it at scale.
Duolingo: Generative Scale and Shifting Interests
The 2025 Annual Report from Duolingo provides the clearest evidence of how generative AI has revolutionized content production. In 2025 alone, the platform launched 148 new language courses. To put this in perspective, it took the company over a decade to build its first 100 courses. With AI-driven tools now handling the heavy lifting of curriculum design and exercise generation, Duolingo has increased its production velocity by an order of magnitude.
The Asian Language Boom The data also reveals a fascinating cultural shift. While English remains the undisputed global hegemon (ranking #1 in 154 countries), interest in Asian languages is exploding. Japanese has surpassed German to become the 4th most popular language globally, and Korean has overtaken Italian to take 6th place. This isn’t just a trend among English speakers; Duolingo’s AI translation capabilities now allow speakers of over 20 other native languages to learn Japanese and Korean directly, without bridging through English. This “multi-source” capability is democratizing access to Asian culture in Latin America and Europe.
The “Teacher-in-the-Loop” Revolution
Perhaps the most significant trend of late 2025 is the pivot away from student-facing chatbots toward teacher-facing AI tools. The industry has realized that replacing teachers is hard, but augmenting them is profitable.
Otus AI: In the 2025-26 school year, the Otus platform deployed an AI assistant to over 106,000 students. Crucially, this tool is designed for administrators and teachers. It allows them to ask natural language questions of their data, such as “Which students are showing early signs of chronic absenteeism?” or “Who is falling behind in math?” This effectively gives every school principal a data scientist on staff, powered by Amazon Bedrock and AWS infrastructure.
AI Math Coach: Similarly, the “AI Math Coach” released by the All Learners Network focuses on “teacher expertise rather than student-facing interventions.” It provides on-demand pedagogical coaching to teachers, helping them plan lessons and understand student misconceptions. By focusing on the adult in the room, these tools avoid the privacy and safety quagmires of direct-to-child AI.
Emerging Trends: Microlearning and Verification
Beyond AI, 2025 has cemented microlearning as the dominant pedagogical format. With attention spans shrinking, platforms are breaking complex subjects into bite-sized, interactive modules that fit into the “cracks” of a busy day.
Simultaneously, blockchain technology is finally finding a practical use case in credential verification. As AI makes it easier to fake essays and exams, blockchain offers an immutable record of degrees and certificates. This is becoming essential for the digital labor market, ensuring that the “skills” a candidate claims to have were actually earned.
Conclusion
EdTech in 2025 is less flashy but more fundamental than ever before. It is no longer about the “magic” of a chatbot, but about the “scale” of 148 new courses and the “insight” of a data query. As we move into 2026, the question is no longer “will AI be used in schools?” but “how effectively can schools manage the AI infrastructure they now rely on?”
An Edu-Tech Person. #Certified By ICDL Africa
