Did OpenAI Cheat to Solve a 90-Year-Old Math Problem? & Teens’ Test Scores Are Plummeting

(upbeat music) Good morning, brood daily show, I'm Neil Fryman. - And I'm Toby Owl. - Today, why teen reading scores are falling off a cliff. - Then, OpenAI solved a famous math problem, but did they cheat to do so? It's Wednesday, September 9th, let's ride. (upbeat music) - Monday might have been Labor Day, but today is another kind of Labor Day. That's because September 9th is the most popular birthday for people in the United States, according to research from 538. They looked at the registered births in the US from 1994 to 2014 and found that September was the most popular birth month and the ninth day of that month today was the most popular birth day. Doing a little back of the napkin biological math. Babies that are born in early September were conceived in late December, which checks out. So I went through and found some famous people born today and I got to say would make for a heck of a dinner party. You got Adam Sandler, you got Hugh Grant, Michael Bubley, then you get Leo Tolstoy and my favorite, Colonel Sanders. Sounds like a pretty fun group to hang out with. My brother's born in September, all these September birthdays. Too many people being conceived around the holiday period, a lot of people inside, a lot of bad weather. I understand why it's happening. Just September birthdays arise up. And our word from our sponsor Hotels.com. Neil, when you stay at a hotel, must have continental breakfast. If I don't start my day with a waffle, it's ruined. Sounds like you could use business preferences from Hotels.com. It helps travelers find the right hotels faster by remembering their go-to work trip requirements. Travelers set their preferences once, including fully responsible stays, Wi-Fi, breakfast, gym access, and more. Then future business trips searches automatically surface matching hotels. That way you can book work trips faster. Just set your preferences once and get results tailored to your business travel needs. Plus get up to 20% off select hotels instantly with member discounts and earn rewards for future travel. Sign up for their free rewards program and start earning on every work stay at Hotels.com/morningbrew. That's Hotels.com/morningbrew. Loyalty members 18+. There are seven problems in the world of mathematics that are equivalent to winning the Super Bowl and World Series at the same time. Established in 2000, they are called Millennium Problems and carry with them a $1 million dollar prize if you can solve them. In the last 26 years, only one has been cracked, but open AI thinks it just solved another. Called the Navier Stokes Problem, it seeks to describe how liquids and gases move, which underpins things like weather forecasting, airplane wings, and even blood flow. The main question mathematicians were trying to solve is does the fluid always remain well behaved or can it get riled up to the point where it does something impossible like water speeding up forever and even blowing up? Open AI says yes, it can. It found a scenario where quote, "In initially smooth fluid at rest can develop a singularity in a finite time." For mathematicians, it's a major finding, a literal once-in-millennium breakthrough. For us, normies, the answer itself is secondary to how it got there. Open AI released a swarm of agents on Navier Stokes, 100 of them initially, but after it was clear progress was being made, the numbers swelled to 10,000. In total, those agents exchanged 2.7 million messages. The total time it spent working was 88 hours and the total compute costs in the millions of dollars all by an unreleased model still in testing. But the funniest part is the reason Open AI sicked a swarm on the problems is because it had caught when that anthropic had cracked two other millennium problems and didn't want to be left behind. We were curious if ours could do it too, Sam Altman said. Neil, this story has it all. Crazy math, a little jealousy between labs and apparently some plagiarism claims for good measure. This has all blown up in Open AI's face from a PR perspective and devolved into a messy, very intriguing drama involving Open AI, perhaps playing dirty here and who should actually get credit for solving the problem. I look forward to watching the Nathan Field or documentary one day because I do not think math gets more spicy than this. Okay, here's what happened by its own admission. As you said, Open AI heard rumors that anthropic may have solved two of these problems and wanting to claim a triumph of its own poured tons of resources and effort into solving Navier Stokes. Turns out the people who were working on it were a professor at NYU named Tristan Buckmaster and a researcher from Anthropic Levent Alpaguey. These guys were close to solving a similar problem to Navier Stokes and had primarily used Open AI's Codex model to help them. Buckmaster got in touch with Open AI and asked them whether any of the data his team had put into Codex was used in the solution to Navier Stokes. In essence, he asked whether Open AI copied their homework. Open AI denied that any specific user data was used to solve the problem, but it couldn't rule out that anonymous data helped train its models. This all left a bad taste in Buckmaster's mouth, not helped by the fact that he also accused Open AI of asking him to leave off his teammates name. Remember, they anthropic guy on a proposal to co-publish alongside Open AI. The company denies this also. Okay, so that's a lot, but the TLDR is Open AI is now seen by many as a bully who essentially stole a human academics painstaking work to solve this $1 million problem to scoop him and score points against his corporate rival. I mean, Almond says it's all overblown. Now that we can see their work, he said the approaches appear to be different, it does raise underlying data questions. So Open AI says it does not train on data from paying enterprise customers. Also, you can opt out of having their data used for trading, but a lot of people say you can just get around that pretty easily by transforming whatever the material data was into synthetic data. And then you can train on that. So that is the big question that people are asking. Just zooming out here, this is a massive breakthrough. Like, I can't overstate how big these millennium problems are. They're literally called millennium problems. This solution still needs to be verified by others in the math community. So it's not just like they checked it off. Some people still think that there actually isn't an answer in here. And they just found a very niche case disproving it. But the other twist is that we are having these harrowing threads emerging from the AI community right now. In addition to this, like once in a generation breakthrough, we had Jacob Coxen, who's a researcher at Anthropic, who quit and says he doesn't want to participate in the industry-wide rush to build systems that can improve themselves and potentially impair humanity. He wrote, "The people building AI earnestly believe that it could kill us all by the end of the decade. This is not a marketing stunt. If anything, many execs in senior researchers will couch their phrasing in the press to sound sensible." So that was one thing. And then that tweet thread that went very viral got quote tweeted by the alignment science lead at Anthropic saying, "Jacob is correct here. We really do earnestly believe AI could kill all humans. I personally think it is less than 10% within the next decade. I believe Anthropic is trying its best." A one-two punster, that is just crazy. He said it was greater than 10%. Greater than 10% within the next decade. I got the little sign wrong. But we do not have a plan to solve alignment. It is just nuts that we had one, this once-in-generation math problem being solved. On the same day that these two Anthropic researchers were basically saying this is superintelligence coming to kill us all. It just goes to show how fast AI progress is both in the positive direction and potentially in a very negative direction too. I mean if Anthropics PR team didn't stay up all night dealing with the fallout of this, I think this is going to be a huge deal. These researchers coming out and saying that Anthropic and OpenAI are not paying attention to these risks at all. And one guy who's still working at Anthropics says there's a greater than 10% chance that AI could kill humans within 10 years. Remember, OpenAI and Anthropic want to go public in the next few weeks or months at $2 trillion valuations. This AI infrastructure build out is basically propping up the entire economy which investor is going to put their money into these companies whose own employees and leaders are saying there's a non-zero in fact there's a greater than 10% chance that this will lead to our own demise in less than 10 years. It is shocking and I think we're going to hear from Washington DC and more politicians about this in the coming days about applying the brakes because you have these companies racing to go public propping up the entire economy and they're also saying that human extinction is possible in not too long. - But at least we solved a cool math problem. - At least we solved a cool math problem. Canada has officially taken off the gloves in its trade brawl with the United States. Yesterday the country imposed retaliatory tariffs of 15 to 50% on $20 billion worth of American goods including cheese, motorcycles, honey, electronics, fish, and cosmetics. If it fills a pledge by Prime Minister Mark Kerney who said he'd respond to US aggression with dollar for dollar tariffs on American items. Several weeks ago trade negotiations broke down at the buzzer and President Trump slapped 50% tariffs on $20 billion worth of Canadian goods accusing Canada of mistreating American products like taking jack Daniels off of shelves. In the interim relations between the once-chummy neighbors have broken down even further with middle school name calling and Trump out of spite renaming Lake Ontario Lake America. Caught in the crossfire are the thousands of businesses on both sides of the border that will face higher costs because of the tariffs potentially passing them on to- consumers or coping by slashing their own work forces. In the deeply integrated auto supply chain, for example, one component could traverse the US-Canada border up to eight times before its final assembly, triggering tariffs each time it passes customs. Analysts believe economic and political pressure in both countries could lead them back to the negotiating table, but for now, they are happy to trade blows. - Trump is really concerned with one company in particular in this escalating trade war as well. He posted on true social, no more selling bombardier in the United States. And bombardier is a Canadian aircraft manufacturer. And it's a perfect encapsulation of this entire trade war because bombardier fired back at Trump and said, "Hey, we actually build a lot in America already." They said they have 3,500 American employees. They have sites in 10 states. They have a supply chain that uses roughly 2,800 American companies across 47 states. Basically, they fired back and said, "We are as integrated with America as can possibly be." It makes no sense to attack us just because we are Canadian in name, but it is really just emblematic of everything that is going on here. It doesn't have to make logical or economic sense, but it just has to be a Canadian company in name and that's why Trump goes after it. - And score political points in the process. This is not huge tariffs. So when you're talking about Canada and the United States trade hundreds of billions of dollars worth of goods and each has slapped $20 billion worth of tariffs on each other, these only account for about 5% of Canada's annual goods exports to the US. However, it does appear that the economic toll will only grow over time. And these particular tariffs that the US has slapped on Canada will disproportionately affect small and medium-sized businesses. Bloomberg profiled one such company, fine cotton factory in Toronto. They export 30 to 50% of their products to the US. And since the US tariffs kicked in, the leader there said, told Bloomberg that, "I've seen a drying up in my orders. I'm probably gonna need to let go of some of my 250 employees as soon as this month because business is just so much slower. This is going to happen to a lot of small businesses on the north side of this border. This is going to happen to a lot of businesses. On the southern side of the border in places like Michigan and Ohio, you have to wonder whether the economic pressure, the political pressure will build on both Mark Carney and Donald Trump to finally reach an agreement because this is going to be painful for a lot of people." - Yeah, the date to circle is January 1st. That is when Trump has threatened to raise tariffs on Canadian metals and vehicles to 50%. That threat is looming over all of these negotiations. Moving on, it's a good thing that Morning Brew has a podcast now, not just a newsletter because the reading skills of teenagers has deteriorated rapidly over the past decade according to a new OECD report. The OECD is a group of 38 mostly wealthy countries that, among other things, test the academic skills of 15-year-olds across the world every three years. They are called the PISA tests and were taken by 760,000 teens across 91 countries last year. The results are in and they are, in a word, bad. Reading skills in OECD countries drop 28 points over the last decade. For context, a 20-point drop is equivalent to losing about a year of schooling. So the drop has been worse than if kids had just stayed home. Math wasn't much better with a 22-point decline while science saw a seven-point drop. But reading is the biggest red flag because in addition to showing the biggest falloff, it's also the most important thing for a kid to learn. The OECD calls it the mother of all skills because the better you are at reading, the easier it is to excel at basically every other subject. What's behind the big drops across the board? OECD education director Andreas Schleiker thinks it's because students have become increasingly hasty when reading. They're more likely to answer quickly than actually reflect on what they read. If an obnox Schleiker blamed on increased screen use, what's on those screens might be an even bigger issue. Nearly half of students use AI chatbots, at least once a week, to help them learn with Schleiker warning that technology is eroding precisely the kind of capabilities it seeks to build. You know, it's just one to test, but piece of results also correlate with economic outcomes for countries. So to see these numbers drop like this, it's not good. - Looking at the country by country breakdown is also interesting. And what you see there is that American students are being absolutely lapped by parts of China and Singapore, Tracy Burns, who's the chief of global strategy and research for the National Center on Education and the Economy. It's estimated that the average student tested in China was more than five years ahead of American students in math and three years ahead in science. When you're looking at math, just 8% of American teens scored at the highest level on the pizza compared with 37% of Singapore, 23% in Korea, 22% in Japan and China, which people are a little critical of their results because they just draw from four particular provinces, more wealthy provinces on the coast, 54% of students scored at the top level in math. So it's not the biggest sample in China, but when you're looking at those four provinces, along with Singapore, along with other Southeast Asian countries, they completely trounce the United States when it comes to learning outcomes. - So I mentioned phones in social media as part of maybe the issue why we've seen this collapse and deep reading, but there's some other explanations too. One is the pandemic is still probably hurting kids. A lot of the 15 year olds today were around 10 when COVID hit, so it's possible that the lingering effects of losing a big chunk of that in-person schooling time is being manifested a few years later. And then another explanation is schools themselves may be getting worse. It's like our worries that schools are increasingly telling students and parents that mediocre work is great. He's saying that schools have become crash test dummies for unproven technologies. So maybe it's just the fact that kids aren't learning because they're not being taught very well. And specifically, Finland may have contributed to this issue. Finland back around the turn of the century was a pizza superstar when it was debuted in 2000. Everyone was shocked that Finland came in at number one. And so everyone went to Finland and say, "What are they doing here that are making these great outcomes come to pass?" And they saw limited standardized testing, limited inspections, more play-based learning, less teacher-led instruction, basically kind of a free for all kumbaya sort of environment. And yet, ever since pizza ranked them at number one, Finland has been on a strong decline, it's basically bottomed out in the rankings. And so people think that they took the wrong learnings from Finland. Actually, what they were seeing was the legacy much more stricter schooling environment was creating great test results. Not this very new age kumbaya approach. And the fact that a lot of Western countries as copy Finland has led them to worse education outcomes, very fascinating how these decisions 10, 20 years ago are now being manifested in all these reading and mass scores falling off a cliff today. All right, we're gonna take a quick break and come back with a story about Singapore right after this. You may have heard about built as the loyalty program that lets you earn points on rent wherever you live. Well, they just leveled up even more. As of 2026, homeowners can also earn up to 1.25x points on their mortgage payments. This is thanks to built three new credit cards, the palladium card, obsidian card, and blue card. All three turn your housing payments rent or mortgage into flexible rewards so you can choose the card that fits your lifestyle without missing out on points and exclusive benefits. Built points can be redeemed at top airlines and hotels, Amazon.com purchases, future rent payments and more. Built points have also been ranked by top publications as the industry's most valuable point currency. Your housing payment is already your biggest expense. Make it your most rewarding. Find the card that fits your lifestyle and apply today at joinbilt.com/mbd. That's j-o-i-n-b-i-l-t.com/mbd. Make sure to use our URL so they know we sent you. Terms and limitations apply subject to approval and eligibility. Built cards are issued by column NA member, FDIC, pursuant to license from MasterCard International Incorporated. Meet Amazon ads, Rising Star, Daryl Spencer, who set out to redefine men's self-care with his brand Crown Skin. Amazon ads helped him grow. I walked away from Big Tech to build a luxury brand in a category that no one was planning attention to. With the help of Amazon ads, I'm reaching relevant audiences and competing with the biggest names and men's grooming. We scaled to over $500,000 in monthly revenue. Watch Crown Skin's story at advertising.amazon.com/RisingDashStars. Every day, shareholders meet to discuss important matters about the company's you invest in. Now, you can easily make your voice heard too. Vanguard investor choice makes it easy to set your proxy voting preference for your eligible Vanguard index funds. In just a few clicks, you can have a say on important shareholder topics like executive pay and director elections. Visit vanguard.com/investorchoice to learn more. It's your shares, it's your voice, it's easy. Vanguard investors own shares of their index funds and those funds own shares of the companies they invest in. Vanguard Marketing Corporation Distributor. Singapore's leader already makes the most of any politician in the world, and now he's getting a big raise. Lawrence Wong will get a $1.1 million boost to put his annual salary at $2.8 million. The most of any world leader by a long shot. The raises part of a broader pay hike for all Singapore ministers, which the government said is necessary to attract better talent and reduce corruption. Wong said, "Overall, I believe this will give me and future Prime Ministers a better chance of persuading capable Singaporeans to step forward and of building the strongest possible team for Singapore." Wong did acknowledge it's a bad look for him to get a raise while regular Singaporeans are concerned about inflation and job security and pledge the entirety of his pay bump to charity. He said, "The sums involved are more than what most citizens earn, so I understand why Singaporeans scrutinize them closely and why many feel strongly about the matter. They do feel strongly about the matter. Back in 2012, backlash against high minister pay was so widespread that they all took a 36% wage haircut. This marks the first pay raise for politicians since that year, and the backlash has returned. Many citizens sounded off against the move on social media, calling it tone deaf, while others said it was a clever play to get smarter people running the government. Toby, no one denies that Singapore's politicians make a lot of money. The question is, is this an effective strategy?" Yeah, they definitely argue that if you want the best people running the country, you've got to compete with the private sector, and actually, since 1994, ministerial salaries have been pegged to the private sector. They look at what the top corporate earners make and try to be competitive there. That pay pegged, though, has actually been ignored because you are right. It's been so politically fraught. Again, you have two schools of thought here. Is that one? You do want to have the very best of the best, and so you have to pay them. That makes sense, but then you are a normal person looking at these ministers, making so much more than you, and you go, of course, I'm going to be mad at the government. Of course, these are the people making the rules. I'm going to rebel against that, which is why you have seen these pay raises be on ice for a long time. The one maybe caveat is that this pay is partly performance-based as well. It's basically like an athletic contract where you have certain thresholds you need to meet where you look at economic growth, you look at unemployment numbers, you look at improvement in the lives of ordinary Singaporeans. That is what feeds into these massive pay packages. It's a very interesting strategy. Does it always play out? That kind of remains to be seen. Yeah, here's how this checks out with the rest of the world. The number two politician pay guy is Hong Kong's chief executive, John Lee, who earns a real chief executive salary of $719,000. Up next is Switzerland's president making $606,000 a year. Then you go down the list to get to Donald Trump, who earns $400,000 a year as president of the United States. Frederick Mertz, the chancellor of Germany, $557,000, and then British Prime Minister Andy Burnham takes in just $230,000. When you look at the scope of what these politicians earn, there's everybody and then there's Singapore. They really do their own thing differently. The big question here is does the high pay actually prevent corruption? Because that is the number one thing you want to hold up is say, pay these politicians enough money so they don't go and look for a little extra cash on the side. Singapore has mostly escaped some corruption scandals, but they're not non-existent. If you go back to just 2024, a former transport minister was jailed for obstructing justice and receiving more than $300,000 worth of gifts. So at the end of the day, I guess they're still humans in the government there. They're still susceptible to a little extra cash under the table, but the general idea is let's pay them a lot so they do not fall victim to corruption. Let's bring to the finish with some final headlines. France's art museums are having trouble protecting their art. Another major heist took place early yesterday morning when thieves broke into a museum featuring Renoir paintings and made off with two. Well, they actually nabbed four, but in their scramble to get out, they left two behind. Together, the four stolen paintings were the most prized of the entire collection. The mayor of the town in southern France, home to the Renoir museum, said the thieves were quote, "well informed, they were well equipped, they did it very quickly and very well." He's right. Currently, the art burglars are outwitting the art defenders. Almost one year ago, thieves broke into the Louvre and stole works worth more than $100 million. In the ensuing months, a number of art museums across Europe, including France, Spain, and Italy, have been targeted by high-profile thefts, highlighting major security lapses for the continent's cultural institutions. Toby, I think they need to call it fluffy. Art Crime historian Laura Evans says that Renoir is actually one of the most frequently stolen artists. It's a very fascinating reason why it's. thefts have been reported in 1951, 85, 1990, 2000, March 2026, and now September 2026. And the reason why is because a lot of his works are very small and portable and also instantly recognizable, which, by the way, is a bad thing for an art stealer. If you steal a piece of art, you don't want it to be essentially something that art dealers can and recognize immediately on site. It makes it very difficult to sell onto the legitimate market because people go, wait, I know that painting, it's been hanging in a museum in France for a long time. So almost all of the other Renoir stolen over the past few decades have eventually been recovered because of that issue. So even though these thieves were very well prepared and, you know, smashing grabbed them under five minutes, they probably won't be able to move the product. They actually stole because it is too dang recognizable. Moving on, Lave Golf spent billions trying to disrupt professional golf. Now it is bankrupt. The Saudi-backed golf league filed for Chapter 11 yesterday with liabilities in the $500 million to $1 billion range, leaving the stars recruited from the PGA tour in Limbo. According to the filing, John Rom is the player owed the most money with a claim of $7.5 million. Major winners Bryson D. Shembo, Dustin Johnson, and Cameron Smith follow close behind. The league has been on life support ever since its main backer, Saudi Arabia's public investment fund, pulled its funding at the reportedly sinking $5 billion into the project. The league says it has found a new investor in the B.C. partners, but a lot of uncertainty still surrounds 2027, which it says will feature a slim-down 10 event season. Neil, instead of Lave Golf killing the PGA tour or joining it, golf's supposed disruptor is now trying to find its way back onto steadier ground. "I don't remember the last time I looked at a bankruptcy report and there's always a list of the top creditors." It's like a leaderboard for a major. John Rom and all these other golfers, the big question is what's the PGA tour is going to do and how it will receive or not receive these golfers who have defected or whether these top golfers, let's say these are big golfers, big draws, how are they going to pursue this path for whether they're going to stay with Liv 2.0 or they're going to try to get back to the PGA tour, so that's a huge question. But I think for a lot of golf fans, there is a lot of shot and Freud about seeing Liv Golf collapse like this. We close out every Wednesday show with suggestion box where Toby and I share a recommendation to get you over at the hump of the week. My recommendation is a book that seems like it's everywhere right now, Lonesome Dove. It lives up to the hype folks released in 1985 by Larry McMurtry. The book tells the story of a cattle drive from Lonesome Dove, Texas to Montana, and is regarded as one of the best westerns ever written, a true masterpiece. It's not short, at 850 pages, but I'm currently on page 820 and dreading reaching the final chapter. The best part is you don't have to leave Lonesome Dove after you finish. There's an acclaimed mini series from 1989 starring Robert DuVal in Tommy Lee Jones, plus more books in the same universe from the author, and it was just announced yesterday that Netflix is developing a new TV series adapting the novel, so just make sure you read that before it comes out. You have been dropping Lonesome Dove left and right, and to see Netflix pick it up, really validated your Lonesome Dove propaganda. I've also asked you, or someone else asked you like, does it sound enjoyable? Would you want to be a cattle driver and you go, no, it sounds off, I could never do it. I don't understand the appeal because I haven't read it yet because it sounds like just a miserable cattle drive over thousands of miles, but clearly it's resonating with people and I'll give it a go. My recommendation is to incorporate a day of whimsy at the gym. A lot of people have regimented gym schedules. Today is back, today is lay, today is chess, but my recommendation, which I definitely saw somewhere, but I can't find exactly where I saw it, is to reserve one day to do whatever you want, whatever machine it calls you, go do it. If you're looking out the pull-up bar and it's calling your name, go try those out. The adductor machine, your foe has been too scared to try, jump on that thing, go nuts. It's supposed to be fun and then, too, it does add a little variation to your gym routine. You might find something that you actually like. I can do this in other aspects of your life, too. You can have a day of whimsy in the kitchen, you can have a day of whimsy walking to work. The day of whimsy is my general recommendation. Who looks at a pull-up bar and goes, oh, so whimsical, I'm just going to take it easy today and have fun. Here's a pull-up bar in front of me, I'm going to do one of the hardest workouts known to man coughing. Maybe there's other things in the gym, but there's always been that one machine that you've walked by and go, huh, I wonder what that one does, and just giving yourself permission to not always be optimized to the max, especially in the gym. I do think it's just good for the mental health, and then also just good for your life in general. All right, that is all the time we have. Thanks for starting your morning with us and have a wonderful Wednesday to share your thoughts on the episode or anything else, send an email to [email protected], or DM us on Instagram @mbdailyshow. Let's roll the credits. Emily Million is our supervising producer. Raymond Lut is our senior producer. Our producer is Olivia Graham, and our associate producer is Olivia Lake. Technical direction by Nina Miller. Heron makeup is sleeping in the morning. And after staying up for that Shelton match last night, Devon Emery is our president and our show is a production of "Borning Brew." - Great show today, Neil. Let's run it back tomorrow. (upbeat music)