A Harvard University PhD candidate has sparked fresh debate online after outlining what she describes as an impending “AI collapse.”
Shae O. Omonijo (@iamshaeo) argues that artificial intelligence systems could soon face serious limitations due to a shortage of high-quality human-generated data.
In a widely shared Instagram post, Omonijo, who studies the intersection of artificial intelligence, the humanities and the future of work, explained that many AI models are at risk of stagnation because they rely heavily on existing human creativity for training.
Omonijo cited predictions from a Wharton Business School professor suggesting that by 2026, AI companies could begin running out of high-quality training data.
According to her, current AI systems primarily learn from human-produced text, images and ideas available online. As more AI-generated content circulates on the internet, models may increasingly be trained on their own outputs rather than fresh human knowledge — a process researchers often refer to as “model collapse.”
“When there’s not enough high-quality human data being produced anymore, models stop growing,” Omonijo said, noting that unlike humans, AI systems do not independently generate new understanding or experiences.
She warned that vast investments in data centers, chips and computing power could be undermined if the underlying data needed to fuel continued improvement becomes scarce.
Despite the warning, Omonijo described herself as an “AI optimist,” arguing that the looming challenge presents an opportunity rather than a dead end.
She pointed to historians, librarians and other humanists as key stakeholders in AI’s future, explaining that enormous amounts of human knowledge remain undigitized in archives, libraries and cultural institutions around the world.
“The people who do have the data are the humanists,” she said, calling for greater respect, better pay and increased funding for digitization and archival projects.
According to Omonijo, bringing this material online could both enrich AI systems and elevate the role of the humanities in the emerging digital economy.
Her remarks come amid growing global conversations about the sustainability of AI development, ethical data use and the balance between technological innovation and human creativity.
As governments, universities and private companies continue to pour resources into artificial intelligence, voices like Omonijo’s are adding urgency to calls for a more inclusive and knowledge-driven approach to the technology’s future.
