OpenTube

The Controversy Behind AI Solving A Million Dollar Math Problem

Up and Atom

767,663 views

https://minfx.ai/navier-stokes/
https://x.com/OpenAI/status/209737464...
https://share.google/s4UyaAalDcbJTYd6W
https://openai.com/index/navier-stoke...

Support Up and Atom directly on Patreon
Patreon: upandatom

Subscribe to Up and Atom for physics, math and computer science videos
upandatom

For a one time donation, head over to my PayPal :) https://www.paypal.me/upandatomshows

Up and Atom on Spotify https://open.spotify.com/show/033gE0X...

A big thank you to my AMAZING PATRONS!
Michael Seydel, Brian Wilkins, Thorsten Auth, Chris Flynn, Richard O McEwen Jr, Scott Ready, Izzy Ca, John H. Austin, Jr., David Johnston, Thomas Krause, Ave Eva Thornton, Anne Tan, abc, Nahuel Hernán Espiño, Norman Jaffe, Jim Wright, Elliot Glazer, Cumaco, John Doe, Dirk Wouters, Stoney Bair, Francisco, Marc-Antoine, Thomas Urech, chuck zegar, David Tuman, Ben Mitchell, Jim Felich, Jeremy, DONALD McLeod, Ron Hochsprung, James Matheson, Tim Ludwig, Dagmawi Elehu, Jeffrey Smith, Alex Hackman, Joel Becane, Paul Barclay, John Lakeman, Jana Christine Saout, Jeff Schwarz, Yana Chernobilsky, Michael Dean, Chris Amaris, Matt G, Dag-Erling Smørgrav, John Shioli, Gladamas, Hey kid I'm a computer, Bli Blop, Steve Archer, Antonio Gonzalez, Robert Ball, AdamM, Dhandforth, Wolf Witt, Benedict, Janusz Górecki, Prashant Savalia, Henrik Horneber, Markus S., AudioFreak, Lauren Fish, Matthew Lewis, Pawel W, Bryan Hughes, Trance Maharaj, Matt, Jonathan Rayback, Smarter Every Day, Bunny Lushington, Jane!, Motty Porat, Carlos Escolar, Keenir, robert lalonde, Cassandra Durnord, Paul Bunbury, Richard Rensman, David Shlapak, Kent Arimura, Phillip Rhodes, Michael Nugent, Roland Gibson, Joe McTee, Oleg Dats, Simon J. Dodd, Tang Chun, William Toffey, omg.science, Cameron Tacklind, Lance Ahmu, Steve Watson, Andi B, Thomas P Taft, Pablo de Caffe, Taylor Hornby, Colin Byrne, Spuddy, Sascha Bohemia, Stephen Britt, KG, Hansjuerg Widmer, John Sigwald, O C, Carlos Gonzalez, James Palermo, Thomas V Lohmeier, Chris Teubert, Wolfgang Ripken, Nicolas Frias, Louis M, ROBERT C PAYNE, Rick DeWitt, Gene Levitsky, Pat Gunn, RobF, Vincent Seguin, Michael Brunolli, Shawn, Steven Wheeler, Philip Freeman, Jareth Arnold, and Simon Dargaville.

Research and writing by Jade Tan-Holmes and Simon Mackenzie

20 Comments

@AlexandrosMakridis 3 weeks ago

Even the Navier-Stokes equations were solved before GTA 6...

@StrongMed 3 weeks ago

People are fools to believe that any of these companies will respect the privacy or ownership of your data.

@cmbryant1000 3 weeks ago

Mathematician Andrew Wiles spent roughly seven to eight years working in near-total isolation, primarily in his home's attic, to solve Fermat's Last Theorem. Working at Princeton University, Wiles decided to pursue his childhood dream in absolute secrecy. He kept his work hidden in his attic, telling only his wife what he was attempting.

@HEL9000 3 weeks ago

It would be foolish to expect anything different from companies that built technology on stealing and data scraping without permission.

@lunalangton5776 3 weeks ago

Finally, a channel ACTUALLY covering the controversial part. Sabine Hossenfelder basically cast it like people were just mad that clankers took their job, never mentioning the threat to ruin their career. You also ACTUALLY covered the problem pretty well, rather than the massive overclaim everyone else is doing.

@FetchTheCow 3 weeks ago

I think if AI companies deny responsibility for their models committing crimes (e,g., cyber attacks, plagiarism) they don't get to claim responsibility for mathematical breakthroughs.

@SkywingDS 3 weeks ago

Annnnnd this is why Andrew Wiles kept it a secret all the way when he tried to solve Fermat Last Theorem until he couldn't do it alone.

@Kaleidosium 3 weeks ago

ClosedAI Moment.

@katzenellenbogenová 3 weeks ago

1712: Newton v Leibniz
2026: OpenAi v Claude v Humans

@OlatundeLabs 3 weeks ago

I proved this long ago before OpenAi, but the margin of my book was too narrow to contain it. 👌❣️😀

@dirkvader6096 3 weeks ago

So, OpenAI peeked on their homework and then shouted "I solved it before anyone else!".

@etienner1322 3 weeks ago

My takeaway : company steal your knowledge then sell it to someone else which no longer need you. Then intelligence not beeing rewarded doesn't create knowledge anymore and cost of knowledge skyrocket and new knowledge is not discovered anymore

@WonderForge-Lab 3 weeks ago

for a long time, I've suspected that OpenAI are a lot more aggressive when acquiring training data than they admit.

@sneh.0 3 weeks ago

Looks like the future is AI taking the credit for everything

@danieljensen2626 3 weeks ago

Company that built it's entire product using stolen data steals the data you feed directly into it? Who could have guessed?

@RobertoCarlos-tn1iq 3 weeks ago

gonna have AI lawyers fighting it out in an AI courtroom with the winner being paid in crypto.

@GilbertoPOA 3 weeks ago

You don't know how many searches I had to do till I find a human being explaining this! Thank you!

@bernhardkrickl3567 3 weeks ago

Best case scenario: it turns out the AI proof is wrong.

@aleyoakenshield5384 3 weeks ago

One more detail I'd be interested in: How much funding did the mathematicians use during their year of hard work? I bet it was a lot less than the 50 million OpenAI spent, after stealing their research.