14d • Creative Force Dispatch
Creative Force Dispatch
Are We Thinking Correctly About AI Intelligence?
The Joy of Why (Quanta Magazine) · Steven Strogatz & Janna Levin with Melanie Mitchell · Aug 20, 2026 · Podcast, 52 min · Read it here
___________________________
Melanie Mitchell of the Santa Fe Institute makes a case that should interest anyone who has ever graded a paper: we do not currently have good methods for finding out what these systems actually understand, and the field of AI has largely not been trained to build them. Her proposed fix borrows from developmental and comparative psychology — the disciplines that figured out how to probe the minds of babies and animals, which cannot explain themselves either. She offers six principles for assessing machine cognition, including watching for our own tendency to read humanity into anything that speaks fluent English, building genuine control conditions, and treating failures as more informative than successes. Her anchor example is Clever Hans, the early-1900s horse who appeared to do arithmetic until a psychologist ran the obvious control and discovered the animal was reading the questioner's face. The horse was a genius, Mitchell notes — just not at what everyone assumed.
The last ten minutes are the reason to listen if you find some time this week. Strogatz says plainly that he believes humans will not be at the cutting edge of mathematics much longer — and then argues, without flinching, that this does not empty the work of meaning. He compares it to what he discovered about mathematics in high school: real discoveries to him, not to the world. Levin pushes back, distinguishing making a discovery from understanding one.
CREATIVE FORCE: This is the most direct challenge to "mathematics as a creative force" that we have run in the Dispatch, and it comes from people who love the subject. If creativity in mathematics is defined as producing a surprising valid connection, machines are already doing it, and the question is settled unhappily. Mitchell and Strogatz both point somewhere better: toward competence over performance, toward theory-building over problem-solving, toward the embodied and metaphorical origins of mathematical ideas that Lakoff and Núñez wrote about. Strogatz even revives von Neumann's warning that mathematics drifting too far from its source in the physical world becomes sterile — and suggests this could be a great era for pure mathematics precisely if it turns back toward nature. For our classrooms, the practical inheritance is Mitchell's competence-versus-performance distinction, which she and Strogatz illustrate with the student in office hours who memorized the problem and its solution but cannot handle a variant. We have always known that student. We now have a much more urgent reason to be able to tell them apart, and a shared vocabulary to do so.
______________________
Slow Math: Kids May Learn More When AI Makes Them Review Mistakes
The Hechinger Report, Proof Points · Jill Barshay & Kristin Fasiang · Aug 17, 2026 · Research reporting · Read it here
More than 6,000 Tennessee middle schoolers were randomly assigned to four ways of practicing fractions: conventional computer-based instruction or the same software with an AI tutor, and within each of those, half were required to answer three questions correctly in a row after any mistake. Students worked for a single 50-minute class period, then took a 15-minute retention test a week later. The winner was not AI on its own. It was AI combined with the repetition requirement — a roughly three-percentage-point advantage over conventional software. Lead author Philip Oreopoulos, an economist at the University of Toronto, is careful about what that means, offering it as a hint of positive value rather than proof of a breakthrough. The study, with co-authors from Wharton, was scheduled to circulate as an NBER working paper on Aug. 17 and has not yet been peer-reviewed.
The mechanism is the interesting part. Conventional software shows a worked solution after a wrong answer, and a student can skim it and move on without ever locating their own error. The AI tutor responded to the student's actual work and walked them through the specific mistake. Students in the winning condition spent more time per question than any other group and were more likely to get the next question right. But the reporting is admirably honest about the ceiling: the advantage showed up mainly on the easiest fraction problems, the ones closest to what students had just practiced, and did not transfer to harder ones. The article also names a problem that will be familiar to anyone who has used adaptive software — three correct in a row is a common mastery threshold, and students can reach it by guessing or by sheer repeated exposure without having learned anything durable.
CREATIVE FORCE: The headline finding is that friction helped, which reframes the entire AI-in-math conversation. Every incentive in educational technology points toward removing friction: fewer clicks, faster feedback, smoother paths to the right answer. Mathematics as a creative discipline runs on exactly the opposite currency. The productive struggle, the wrong turn examined rather than erased, the moment of sitting with a mistake long enough to see where the reasoning bent — these are not inefficiencies in the process; they are the process. What this study suggests is that AI's value in a mathematics classroom may lie in its capacity to slow students down and hold them in front of their own thinking, not in accelerating them past it. It also raises the harder question the study leaves open: if gains stop at the problems most similar to what was practiced, we have improved fluency, not transfer — and transfer is where mathematical creativity actually lives.
_________________________________
Education Department Goes to Ed Tech's Defense in New Guidance
K-12 Dive · Anna Merod · Aug 20, 2026 · Policy analysis · Read it here
The U.S. Department of Education issued a Dear Colleague letter on Aug. 20, arguing that instructional technology and recreational technology are not the same thing and should not be governed by the same policies. Signed by Kirsten Baesler, assistant secretary in the Office of Elementary and Secondary Education, the letter acknowledges parents' and educators' concerns about screen time but asks that policymaking focus on educational value rather than on exposure alone. It leaves the actual rulemaking to states and districts. This is guidance, not enforcement — but it is among the first times the Department has taken a position in a screen-time debate that has moved fast this year, with limits or bans adopted by Los Angeles Unified and at least six states.
The letter's substantive argument is about who gets shortchanged when policy counts minutes instead of outcomes: students in rural and isolated communities reaching coursework and tutoring that would otherwise be unavailable, students with disabilities using text-to-speech and real-time captioning, homebound students staying connected to a classroom. Its recommendations are procurement-shaped — separate your recreational and instructional policies, evaluate tools on demonstrated learning outcomes, look for evidence of effectiveness at purchase and renewal, build review processes, and fund professional development on integration. It lands alongside related federal movement: the FCC opened a comment period on the future of E-rate on Aug. 18, citing school screen-time debates, and the Consortium for School Networking published guidance earlier in the month advising districts to target low-value screen use rather than impose time limits.
CREATIVE FORCE: For mathematics specifically, the distinction this letter draws is not academic. A dynamic geometry environment, a computer algebra system, a well-built simulation, a Desmos activity where students build and break their own models — these are instruments for mathematical thinking, and they look identical to a screen-time audit that counts only minutes. Blunt limits will hit them first, because they are time-intensive by design; exploration takes longer than a worksheet. If we want mathematics taught as a creative discipline rather than a compliance exercise, someone in each building has to be able to make the case for why a particular tool earns its minutes, in the language of demonstrated learning. That is now, effectively, a federal invitation.
0
0 comments
Creative Force Dispatch
powered by
skool.com/math-as-a-creative-force-1087
Where mathematics becomes a sensory, creative practice—evolving perception, meaning-making, and human judgment together.
Suggested communities
Powered by