Coursera CEO: ChatGPT will transform how we evaluate people's skills

Jeff Maggioncalda, the American chief executive of Coursera, arrived in Brazil last week with a conviction that most educators have not yet embraced: artificial intelligence will not destroy education, but fundamentally reshape it. While universities worldwide grapple with how to respond to ChatGPT—some banning it outright, others cautiously experimenting—the leader of one of the world's largest online learning platforms sees the technology as a watershed moment for how we teach and evaluate what people actually know.

Maggionacalda spent his time in Brazil meeting with governors, including Romeu Zema of Minas Gerais and Mauro Mendes of Mato Grosso, to discuss potential partnerships bringing Coursera's courses and credentials to public sector workers. The company, valued at $1.9 billion on the New York Stock Exchange and serving roughly five million registered students in Brazil alone, is not waiting passively for the AI moment to arrive. It is building toward it. In April, Coursera launched what it calls the Coursera Coach—an AI assistant powered by OpenAI technology that functions as a personalized tutor. A student watching a lecture can ask the coach to explain a concept in simpler terms, request a summary, or continue a conversation about material they do not understand. The tool works in any language, even if the original course is in English. By the end of this year, the platform plans to use AI to translate 2,700 of its 5,800 courses into Brazilian Portuguese, expanding from just 280 localized offerings today.

But the real transformation Maggioncalda sees goes deeper than tools and translations. It concerns the very architecture of how education measures competence. For generations, teachers have evaluated students by examining their written work—the essay, the problem-solving steps, the final answer. Writing itself was understood as a window into logical reasoning. ChatGPT has shattered that assumption. A student can now produce polished prose without having thought through the underlying logic at all. This forces a reckoning. Maggioncalda argues that educators will have no choice but to shift their focus from the final product to the thinking process itself. Some professors are already experimenting with this: they ask students to use ChatGPT but then evaluate how the student deployed the tool, what questions they asked, how they refined their prompts. This mirrors methods used at Oxford and Cambridge for centuries—the Socratic method, a conversation between teacher and student designed to expose gaps in understanding. The virtual coach, Maggioncalda suggests, could democratize this approach, making it available to anyone with internet access rather than only to students at elite institutions.

Maggionacalda himself uses ChatGPT as a kind of executive sparring partner. He poses strategic questions, asks how to improve his interview techniques, describes situations and requests feedback on his own critical thinking. His secretary has begun feeding the tool samples of his emails, then using it to draft responses in his voice based on key topics—responses he reviews before sending. It saves time. But this casual efficiency masks a deeper concern that Maggioncalda articulates with measured worry. If people become lazy, they may outsource their thinking entirely to the machine. The act of writing itself is where much thinking happens; without it, people risk atrophying their capacity for critical thought. And this matters most precisely when AI is generating false information at scale, when the ability to think critically becomes more essential, not less.

His moderate fear extends further. Social media algorithms—pure artificial intelligence—evaluate and amplify content based on a single metric: engagement potential. The result, he argues, is the destruction of society. If AI already controls the information ecosystem, what happens when it begins managing energy grids, transportation networks, weapons systems? We already have a preview of what can go wrong. The pandemic accelerated online learning dramatically: Coursera grew from 76 million enrollments in the first year of COVID to 143 million the next year, then to 233 million. The growth has continued, driven partly by recognition that jobs will change and skills will matter more than ever. Remote work has amplified this effect, particularly in emerging markets. Africa, India, and Latin America show the fastest growth—young workforces recognizing that online learning offers access to better-paying jobs that were once geographically out of reach. This is wage arbitrage, Maggioncalda notes, but it is also genuine opportunity. Before, talent was concentrated where institutions existed. Now it is distributed wherever internet exists.

Maggionacalda's visit to Brazil reflects this calculation. He is focused on three areas: workforce development programs to make courses more accessible, teacher training for online instruction, and upskilling of public employees. Espírito Santo recently launched a pilot program offering courses to civil servants, teachers, and citizens. In six months, it attracted more than 12,000 enrollments and nearly 66,000 completed lessons. Other Brazilian states are watching. The question now is whether the education system can move as fast as the technology demands—whether assessment methods can evolve, whether teachers can be retrained, whether the promise of democratized learning can outpace the risks of democratized misinformation.