A Hobbyist, a Chatbot, and an 80-Year-Old Problem
Liam Price has no formal mathematical training and hasn’t yet attended university — but working from home in southwest England, he used ChatGPT to break new ground on Erdős problem #1196, one of more than a thousand puzzles the Hungarian mathematician Paul Erdős collected over his lifetime. What unsettled specialists wasn’t just the answer but the route: where human solvers had reached for the language of probability theory, the model worked in the problem’s original framing and, in doing so, implicitly built a bridge between numbers and probability. The piece sits against a broader backdrop — AI also cracking a separate 80-year-old challenge — and captures a field watching its own methods shift under its feet.
CREATIVE FORCE: When a result can come from someone outside the credentialing system entirely, the question stops being “can AI do mathematics?” and becomes “who gets to be a mathematician, and what counts as doing the work?” That’s a question math educators will be answering for their students very soon.
‘The Job Description Is Changing’: Terence Tao on the Rise of AI
If the Price story was interesting for you, this interview could be considered the sequel as it answerthe questions in the CREATIVE FORCE line. The Fields medalist pushes back on the loudest framing — a wave of coverage has suggested mathematics is being fundamentally overturned, and Tao notes that AI’s real capabilities are often hyped, and that it’s premature to declare the profession dead. But he doesn’t dismiss the shift either: by his account, in just the past year AI has moved from secondary-school-level problems to being genuinely useful in working mathematicians’ daily research. His through-line is collaboration, not replacement — a recalibration of what the day-to-day of the discipline looks like rather than its obituary.
CREATIVE FORCE: Tao is the rare voice trusted by both the hype skeptics and the true believers. For educators, his “collaborator, not replacement” framing is a usable model for talking to students about AI — one that neither panics nor over-promises, and that keeps human judgment at the center of the work. How can AI be a collaborator in learning and inquiry that you foster in your math classes?
We Banned the Phones. The Test Scores Didn’t Move.
Stanford economist Thomas Dee, a co-author, calls this study a first draft of a new way to measure what cell-phone bans actually accomplish — and the headline finding is a paradox. Wellness improved, but the effect on test scores in the first three years after a ban was close to zero: a small positive bump for high-school math, a small negative one for middle schools. The researchers’ candid explanations are the most useful part: students may simply migrate their distraction to laptops and tablets, and any gain from fewer phone interruptions can be offset by the disruption of teachers policing the ban. A clear-eyed look at the distance between an intuitive policy and a measurable result.
CREATIVE FORCE: This is the classroom-tech version of the math stories above: the promise of a technology (or its removal) and its measured effect are not the same thing. For practitioners weighing device policy, the lesson isn’t “bans don’t work” — it’s that the win may be well-being, not scores, and that implementation cost is real. Name the goal you’re actually optimizing for. I believe that these studies also reveal the underlying truth... the teacher in the classroom makes the difference, not the addition or subtraction through policy of tools.