The school is now putting AI versions of its faculty to work in a $699 online startup bootcamp, where aspiring founders can rehearse investor pitches, sales calls, and board meetings—all with AI clones of Harvard instructors. They can even test their pitches with an AI avatar repeatedly before a real meeting.

The eight-week program is part of HBS Foundry, which Harvard describes as an “AI-native digital workspace” for entrepreneurs. The startup bootcamp is equipped with “personalized AI mentorship modeled on Harvard Business School faculty” alongside live sessions with experts, culminating with an opportunity to pitch to investors for $100,000. So far, 760 founders have participated in the bootcamp, according to a Harvard spokesperson.

“The idea for the AI experiences came from founders themselves, who wanted opportunities to practice and be challenged before getting in front of mentors and investors, so those conversations could be even more meaningful,” Kat Rings, Foundry’s director, told Fortune. 

The “clones” include seven HBS professors and senior lecturers with backgrounds spanning venture capital, business-model design, board dynamics and startup strategy. Their participation was voluntary, with the program having dedicated interviews and recording sessions, testing and ongoing faculty feedback as part of the process.

“The goal isn’t to create a substitute for me,” Shikhar Ghosh, one of the cloned HBS professors , told Fortune. “It’s to give founders another way to prepare and challenge their thinking, so that when we do come together live, we can spend more of that valuable time on the questions where human judgment and conversation matter most.”

Notably, Foundry’s startup bootcamp is not an HBS degree program nor for academic credit, but rather, a way to “extend entrepreneurial support to high-potential founders wherever they are, including people who may never study at Harvard or have access to the networks and resources of a major innovation ecosystem,” the Harvard spokesperson told Fortune.

The experiment falls into the growing dichotomy of how AI is addressed and used on campus. On one hand, professors have complained about students cheating on exams using AI, with Brown University’s Robert Serrano, for example, urging universities to stamp out AI-enabled cheating. On the other hand, some universities are using AI to scale otherwise scarce faculty time into a digital product. Imperial Business School has already created digital twins of professors, while Georgia Tech has deployed AI teaching assistant Jill Watson across large online classes. 

Trend of universities playing with AI chatbots in education

Harvard is not the first school to experiment with turning a professor into software. Last year, Georgia Tech professor David Joyner had created an AI avatar of himself, dubbed “DAI-vid,” for an online generative AI course on edX, an open online course provider founded by Harvard and MIT. The avatar was modeled on Joyner’s appearance and voice and was framed as part of a broader effort to democratize online learning.

“Much like online courses have made lectures from top universities broadly available, AI has the potential to go a step further and make not just a professor’s knowledge, but some version of their perspective and judgment, available at scale,” Lindsay Tanne Howe, whose company LogicPrep helps students get into top schools, told Fortune. “But what it can’t replicate is everything that happens between people.”

Kian Katanforoosh, a Stanford lecturer and founder of AI skills platform Workera, told Fortune that AI “can replicate many of the things professors do repeatedly” like answering unlimited questions and explaining concepts in different ways, but can’t truly know a student or be there for them.

“What AI cannot fully replicate is knowing a student as a person,” Katanforoosh said, explaining that a professor can recognize potential, inspire a student and make an introduction based on years of human interaction.“AI can simulate some of these behaviors, but today it cannot put its own reputation at stake or genuinely care what happens next.”

Scaling faculty through AI also creates a potentially awkward question for universities: At what point does efficiency begin to undermine what students believe they are paying for? College students already object to professors outsourcing lesson planning and grading to AI, as Fortune previously reported, and while AI is part of Harvard’s Foundry proposition and not going on behind the scenes, it could raise a similar tension.

“It can’t necessarily replicate the experience of an elite education,” Howe said. 

Joyner, who’s now Georgia Tech’s interim Vice Provost for AI in Education, told Fortune that Harvard’s startup bootcamp’s price tag–relatively inexpensive compared to the business school’s annual tuition of over $84,000–is part of the push to make higher education more available, but stressed that the role of the professors themselves is “key” to making AI avatars. 

“Faculty using AI to expand the number of students that they can reach and the quality of the instruction they give can have an enormous positive effect,” Joyner told Fortune over email. “But if companies or universities try to build these tools out of existing content without strong collaboration with the person being represented, there’s no way to really know if the experience is representing the faculty themselves.”