Why Being Smart Makes You Replaceable | Ken Ono, Axiom Math
In a Nutshell
Ken Ono, a mathematician at Axiom Math, realized AI like ChatGPT outperforms humans in knowledge recall, making raw smarts replaceable—true intelligence lies in judgment, reasoning, pattern recognition across disciplines, and creative inference, not facts. Inspired by Srinivasa Ramanujan's autodidact genius and his own dropout path to expertise in mock theta functions, he critiques high-stakes education for stifling passion and advocates AI as a cheap tutor to foster wonder, curiosity, and self-directed learning over perfectionism and debt-trapping careers. Undiscovered "Ramanujans" exist everywhere; nurture bravery in following interests to propel fields forward.
These notes were generated by AI and may contain inaccuracies.
Exactly one year ago, Ken Ono was a happy-go-lucky university professor writing papers, enjoying the privilege of being good at mathematics. Then came a dramatic change: he became deeply troubled. For the first time, he struggled to assemble questions that ChatGPT would get wrong. These models know more facts than any human. He was devastated, wondering how to stay ahead of AI, but realizes that's the wrong question. Knowledge has become cheap; trying to always stay ahead of AI means losing.
Ken Ono is a mathematician working in AI for math, a professor at the University of Virginia on leave, and the founding mathematician at Axiom Math.
In 1993, the first time he heard 'artificial intelligence,' he met a faculty member and cockily said, "I specialize in natural intelligence." A year ago, as a professor at the University of Virginia, he faced large language models through the Frontier Math program by Epoch AI in Berkeley. They hired professional mathematicians worldwide to create difficult math problems to assess LLMs. He struggled to make questions ChatGPT would get wrong and had early access to state-of-the-art models.
These models know more facts than any human you would ever find.
LLMs are the most extraordinary librarian: if written down, on YouTube, or in newspapers, they've seen it. Nobody watches Usain Bolt race a motorcycle; society accepts machines outperforming humans physically, now intellectually too. Information collection is unbeatable, but using and verifying it is expensive. No one wants a librarian as neurosurgeon or air traffic controller—human judgment matters.
Intelligence is reasoning, making inferences, creating new concepts, generating ideas, stringing concepts deeply—not regurgitating facts. We're bad at teaching this. It's setting dials to design systems from scratch for gadgets, programs, or new science. Recognizing patterns transferable across disciplines to propel fields forward is genius—not just luck. Hard-nosed experts committing daily to their niche are also intelligent and genius-worthy.
Son of a mathematician, considered gifted young; parents planned: oldest brother pianist (succeeded), Ken youngest mathematician, middle son (deemed bad at math/music) bank worker—which drove him to become president of the University of Michigan, a distinguished scientist.
Ken dropped out of high school, rebelling against expectations as the Asian kid good at math, unable to play baseball. In April 1984, planned to run away. A letter arrived for his father from Janaki Ammal, widow of Srinivasa Ramanujan, thanking for a gift toward a statue promised in 1920. Seeing emotionless father cry moved him.
Ramanujan was a mystic autodidact with goddess-given formula visions, flunked college twice due to passion for math alone, left three notebooks. Father studied him; post-WWII Japan, math escaped food lines. US mathematicians rebuilt Japanese universities; father learned of Ramanujan at a conference where Princeton professor discovered him, launching his career. Ramanujan died at 32, nearly forgotten; global mathematicians funded statue.
Ramanujan gave hope: parents' hero was a two-time dropout, not perfect Ivy Leaguer. At University of Chicago, horrible student until PBS documentary on Ramanujan jumpstarted him. Biography The Man Who Knew Infinity followed; he pursued PhD thesis on Ramanujan's work, developing theory of Galois representations to study his mock theta functions. By 1993, Fermat's Last Theorem proof by Andrew Wiles relied on it—best career decision.
Where would we all collectively be had [Ramanujan] not been discovered? A world I cannot fathom. There must be other Ramanujans.
Ran Spirit of Ramanujan program finding undiscovered talent; former student Karina Yong was early recipient. Creative potential resides in all; give students opportunity to bravely follow curiosity in embracing systems.
Top students worldwide stressed over school/college admissions, test scores—checkbox participation is messed up. Participate but recognize it. Education inspires wanting to know the world, cultures. Play for infants is science: stacking blocks teaches gravity through joy.
Students should say, "This is a wonderful class because the subject is beautiful." But system fixates on GPA for grad/med/law school—opportunity lost.
AI transition: it reads papers better than he remembers, answers adjacent math questions without judgment. Knowledge cheap via internet/LLMs ($80,000/year US college vs. self-paced AI learning). Miss human elements: deriving right questions, next frontiers—why college/professors needed. AI excels at tutoring, precision learning.
Kindergarten wonder (knowing primes, adding) should be bottled. Best scientists/doctors view world wondrously, benevolently.
Value perfection/speed in tests fails to train next Einstein. For children, foster passion for world: climate, conflicts, cultures. US education debt: $150k undergrad + $200k professional school traps in hated careers—purgatory. Life becomes bill-paying; own your identity.
Keep EO in your library
Save the videos and channels worth coming back to, and find them again in one place.
