AI Isn't Saving Time — It's Quietly Adding a Second Job

The paper's authors interviewed 36 graduate students who use AI chatbots like ChatGPT for academic work — half of them native English speakers, half not. The puzzle they set out to solve was simple: why do people keep using a tool they openly say is making their writing, thinking, and skills worse? Most studies assume that continued use means people are satisfied. This one argues the opposite.

Instead, the researchers describe a repeating cycle they call Active Negotiation. Something goes wrong — the AI invents a fake citation, or writes in a flat, generic voice — and the user springs into action: fact-checking, rewriting, re-prompting. Then they talk themselves into continuing. The cycle starts over the next time something breaks.

That labor happens in three places. Outwardly, users audit the AI's output. Inwardly, they audit themselves — wondering whether their own thinking is getting weaker. Socially, they manage how classmates and professors see them, sometimes hiding their AI use entirely. Students who speak English as a second language had one more reason to stay: AI puts them on equal footing with native speakers.

The takeaway for everyone else: "using AI" isn't a click, it's a relationship you have to maintain. The time AI saves often comes back as time spent checking it. And competitive pressure — captured in the paper's title, "If you're not doing it, somebody else is" — keeps people locked in. This is early research based on just 36 students, so treat it as a strong signal, not a final verdict.