Pearson's stock still fell about 30 percent over the past year, even as the company grew its business. Investors assumed AI would do to textbook publishers what it already did to homework-help sites. That fear was reasonable. It just turned out to be only half right.
Look at Chegg. Its entire business was explaining textbook concepts and answering homework questions. Once ChatGPT started giving away the same explanations for free, Chegg's stock crashed from a peak of $115 a share to near $1. The company also cut nearly half its staff within a single year. Coursera and Duolingo got similar downgrade warnings from Wall Street analysts, for the same reason: their value depended on delivering content a chatbot can now produce instantly and for nothing.
Pearson looked just as exposed back in 2023, when its market value dropped on the same day Chegg's stock collapsed. But look at where Pearson actually makes its money today. Roughly 80 percent of its profit comes from testing and virtual schools, not from the textbook-style courseware that AI threatens most. Testing and certification work, things like GED prep, professional licensing exams, and accredited online schools, needs something a chatbot cannot fake: legal accountability. Pearson's chief executive points out that this part of the business has to prove its results hold up, sometimes to a court, sometimes to a regulator. That proof is what customers are actually paying for, not the content itself.
This is the real lesson for any business owner watching AI move into their industry, education or otherwise. Ask a simple question: does your revenue come from delivering information, or from vouching for something, a result, a qualification, a compliance record? If it's the first, expect the same pressure that hit Chegg. If it's the second, there is more room to breathe, because AI can write an answer, but it cannot yet stand behind that answer the way a regulated business can.
The claims about AI actually improving how students learn deserve more scrutiny than they usually get. Pearson points to numbers like a large jump in students reaching mastery through adaptive practice. But researchers such as Mutlu Cukurova at University College London note that measuring behavior inside the same platform providing the AI does not prove students learn more overall. Outside research is mixed in the same way. Some trials show AI tutoring beating classroom teaching. Others found it improved homework scores without moving exam results at all. Anyone buying an AI training tool for staff should ask for proof beyond usage statistics, the same gap Pearson admits it has not fully closed.
The deeper threat to Pearson's advantage is not the AI models themselves. It's that content alone stopped being scarce. Once any explanation can be generated instantly, what customers actually pay for shifts toward relationships, verification, and institutional trust, exactly the part of the business that a foundation model still cannot absorb.