The time of avatars in the classroom has arrived: Harvard clones its faculty

David Shrier has not always controlled his image. When this university lecturer left the institution where he had taught, the school continued using his face to promote an online entrepreneurship course with videos he had recorded, and without paying him. He had to hire a lawyer to have them removed from the internet. But the idea did not disappear after the lawsuit. Today Shrier teaches at Imperial Business School in London and he has a digital replica that answers his students’ questions at any time of the day. They call it Davatar. “What I know is that now my name, my image and my appearance belong to me,” he says.

Artificial intelligence (AI) has made it easier for universities to create virtual replicas of their professors. Since April of this year, Harvard Business School has offered HBS Foundry, an eight-week online program priced at $699 aimed at startup founders. The platform projects video avatars of seven professors, created by the U.S. company HeyGen. In front of them, entrepreneurs—graduate and doctoral students—practice investor pitches, sales calls and board meetings. According to The New York Times, Harvard has tested it at more than 100 universities to date. The course is open to the public, although applicants must request access in writing.

Katharina Rings, director of the initiative, says the sense of proximity that AI-driven avatars can generate is one of the program’s keys to success. “We repeatedly hear that AI can seem anonymous, almost like a black box,” she tells EL PAÍS by email. With an avatar, the entrepreneur knows there is a real person behind it, which increases interaction, Rings adds. Foteini Giannopoulou, the program’s instructional designer, says AI opens a new path for business schools, where practice is essential: “With AI we can go beyond access to information and move toward broader access to practice.”

Harvard, however, does not yet have data proving that avatars improve learning. “Foundry launched recently, so we do not yet have longitudinal results comparing founders who practiced with the avatars with those who did not,” Rings acknowledges. Surveys have shown that students who use the program arrive in a calmer state of mind to business meetings, and that investors report seeing them deliver more precise presentations.

Enrique Dans, a professor of innovation at IE University in Spain, also does not take it for granted that these programs improve learning. “They could deliver equivalent outcomes at lower cost or serve people who currently receive no help. But without independent evaluation, we only have a marketing tool with the appearance of an educator.”

Pioneers in the sector

The other pioneering experiment is at Imperial College, where Shrier is a lecturer. He launched his avatar in January 2024 with his AI Ventures students, a course that, by his count, has grown from 78 to about 200 students. The replica is fed more than half a million words from his writings and materials, according to the academic.

One of the biggest fears raised by these initiatives is that AI avatars may invent content not previously authorized by each lecturer. AI tutors have already made mistakes in the past. In February 2024, The Wall Street Journal tested Khanmigo, the tutor from the educational organization Khan Academy. The system failed on basic subtraction and accepted an incorrect answer to a Pythagoras theorem problem with an enthusiastic “Excellent!”

Rings says that at Harvard each professor “can review what their avatar says and make changes at any time.” Jeff Bussgang, a school professor and co-founder of venture capital firm Flybridge Capital, is the program’s most visible face. At first, Bussgang reviewed his replica’s conversations through a portal. Although he admits he no longer does: “I simply trust Katharina, her team and the algorithm,” he confesses.

By contrast, Shikhar Ghosh, another professor with a virtual replica at Harvard, updates his avatar as his thinking evolves and receives a notification each time someone tries to push the virtual simulation beyond his field of expertise. One user once asked it to draft his prenuptial agreement. Shrier, for his part, works with Star Echo, a company founded by a former student, with which he was able to set limits on the content his digital clone may generate. “I don’t want my university’s official avatar to be forced to say anything,” he stresses.

Dissatisfaction

On U.S. campuses, some faculty say AI is being introduced without their participation. In a July 2025 report based on a survey of 500 academics, the American Association of University Professors (AAUP) found that 71% perceive AI’s introduction to be overwhelmingly led by administrators, with little faculty involvement. The document also warns about implications for intellectual property.

Concerns at Imperial College faculty center mainly on intellectual property and the fear that AI will attribute words to them they do not agree with, for example misinformation or outdated information, according to O’Grady, chief learning designer at the university’s Idea Lab. Dans argues that primary responsibility should rest with the university offering the service, without prejudice to the responsibility of the technology provider and the lecturer if they have ceded their identity. “The risk is especially serious because professors’ faces and voices lend any response an assumed authority, meaning the student does not perceive an algorithmic hallucination but a professor’s assertion,” the academic comments.

The United States does not have a federal law on digital replicas, although some states, such as California and Tennessee, already regulate their use. The No Fakes Act, which would lay the groundwork for a national rule, passed the Senate Judiciary Committee in June and is still under consideration. In the European Union, since August 2 of this year, Article 50 of the Artificial Intelligence Regulation requires companies to notify users that they are interacting with an AI and to label content that imitates real people. In Spain, the Council of Ministers approved on July 7 a draft organic law that deems illegitimate the use, without authorization, of someone’s voice or image for advertising or commercial purposes through AI. Despite all this, Bussgang finds no limits to the use of avatars at the university. The format has “universal applicability,” he says. “I can’t think of an environment where it wouldn’t work.”

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition