Google and OpenAI's AI models win milestone gold at global math competition
Alphabet's Google and OpenAI said their artificial-intelligence models won gold medals at a global mathematics competition, signaling a breakthrough in math capabilities in the race to build powerful systems that can rival human intelligence.
The results marked the first time that AI systems crossed the gold-medal scoring threshold at the International Mathematical Olympiad for high-school students. Both companies' models solved five out of six problems, achieving the result using general-purpose 'reasoning' models that processed mathematical concepts using natural language, in contrast to the previous approaches used by AI firms.
The achievement suggests AI is less than a year away from being used by mathematicians to crack unsolved research problems at the frontier of the field, according to Junehyuk Jung, a math professor at Brown University and visiting researcher in Google's DeepMind AI unit.
'I think the moment we can solve hard reasoning problems in natural language will enable the potential for collaboration between AI and mathematicians,' Jung told Reuters.
OpenAI's breakthrough was achieved with a new experimental model centered on massively scaling up 'test-time compute.' This was done by both allowing the model to 'think' for longer periods and deploying parallel computing power to run numerous lines of reasoning simultaneously, according to Noam Brown, researcher at OpenAI. Brown declined to say how much in computing power it cost OpenAI, but called it 'very expensive.'
To OpenAI researchers, it is another clear sign that AI models can command extensive reasoning capabilities that could expand into other areas beyond math.
The optimism is shared by Google researchers, who believe AI models' capabilities can apply to research quandaries in other fields such as physics, said Jung, who won an IMO gold medal as a student in 2003.
Of the 630 students participating in the 66th IMO on the Sunshine Coast in Queensland, Australia, 67 contestants, or about 11 per cent, achieved gold-medal scores.
Google's DeepMind AI unit last year achieved a silver medal score using AI systems specialized for math. This year, Google used a general-purpose model called Gemini Deep Think, a version of which was previously unveiled at its annual developer conference in May.
Unlike previous AI attempts that relied on formal languages and lengthy computation, Google's approach this year operated entirely in natural language and solved the problems within the official 4.5-hour time limit, the company said in a blog post.
OpenAI, which has its own set of reasoning models, similarly built an experimental version for the competition, according to a post by researcher Alexander Wei on social media platform X. He noted that the company does not plan to release anything with this level of math capability for several months.
This year marked the first time the competition coordinated officially with some AI developers, who have for years used prominent math competitions like IMO to test model capabilities. IMO judges certified the results of those companies, including Google, and asked them to publish results on July 28.
'We respected the IMO Board's original request that all AI labs share their results only after the official results had been verified by independent experts and the students had rightly received the acclamation they deserved,' Google DeepMind CEO Demis Hassabis said on X on Monday.
OpenAI, which published its results on Saturday and first claimed gold-medal status, said in an interview that it had permission from an IMO board member to do so after the closing ceremony on Saturday.
The competition on Monday allowed cooperating companies to publish results, Gregor Dolinar, president of IMO's board, told Reuters.
Reporting by Kenrick Cai and Anna Tong in San Francisco and Jaspreet Singh in Bengaluru, and Krystal Hu in New York; Editing by Matthew Lewis.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Globe and Mail
6 minutes ago
- Globe and Mail
META: Liquidity, Growth, and AI Fuel Institutional Appetite
Meta Platforms Company Overview Zacks Ranks #3 (Hold) stock Meta Platforms ( META ) is the world's largest social media social media powerhouses like Instagram and Facebook under its umbrella, the company is benefiting from increasing ad revenue. Additionally, Meta is working on growing and monetizing WhatsApp, the world's largest messaging app. More than 3 billion people use Meta's family of products on a daily basis. Though Meta faces intense competition from Alphabet (GOOGL), Twitter, Amazon ( AMZN ), and Snap ( SNAP ), the company continues to gain market share. Meta: Liquidity + Fast, Consistent Growth = Institutional Appetite With double-digit EPS and revenue growth, it's difficult for investors to find a company growing as fast, consistently, and with such liquidity as Meta. Because META enjoys this rare trifecta, institutional investors are likely to remain attracted to this 'hedge fund hotel.' Meta Benefits from AI Now and into the Future Meta Platforms is expected to spend roughly $70 billion in CAPEX in 2025 to boost its AI ambitions. While the spending may seem reckless to Wall Street due to the sheer size, META is already enjoying significant returns on its AI investments, unlike most AI companies. In Meta's recent earnings conference call, visionary CEO Mark Zuckerberg featured five ways AI will drive META, including: 'Improved advertising, more engaging experiences, business messaging, Meta AI, and AI devices.' AI is driving increased time spent on its core social media offerings. AI optimization has led to a 5% increase in time spent on Facebook and a 6% increase on Instagram – leading to better returns for Meta's ad business (which makes up ~97% of revenue). Meanwhile, Zuckerberg made a shocking forward-looking statement that suggests that his company's AI is the best or near the best, saying: 'Developing superintelligence – which we define as AI that surpasses human intelligence in every way – we think is now in sight.' META: Post Earnings Drift Set-up Last week, META blew away Wall Street EPS expectations, marking the company's eleventh consecutive earnings beat. Following its earnings beat last week, META shares bolted more than 11% as volume swelled to 220% above the norm. Heavy volume and robust price action of this magnitude signal heavy institutional accumulation among investors. Additionally, because these institutional investors usually buy for weeks and months at a time, I expect META to benefit from being a classic 'post-earnings drift' play. Bottom Line Meta is one of the rare companies already benefiting from its AI investments. The company is being driven by strong performance across its social media platforms, with consistent growth in both revenue and earnings per share. Zacks Names #1 Semiconductor Stock This under-the-radar company specializes in semiconductor products that titans like NVIDIA don't build. It's uniquely positioned to take advantage of the next growth stage of this market. And it's just beginning to enter the spotlight, which is exactly where you want to be. With strong earnings growth and an expanding customer base, it's positioned to feed the rampant demand for Artificial Intelligence, Machine Learning, and Internet of Things. Global semiconductor manufacturing is projected to explode from $452 billion in 2021 to $971 billion by 2028. See This Stock Now for Free >> Want the latest recommendations from Zacks Investment Research? Today, you can download 7 Best Stocks for the Next 30 Days. Click to get this free report Inc. (AMZN): Free Stock Analysis Report Snap Inc. (SNAP): Free Stock Analysis Report Meta Platforms, Inc. (META): Free Stock Analysis Report
Globe and Mail
2 hours ago
- Globe and Mail
Is MRVL Stock a Buy, Sell or Hold at a P/E Multiple of 7.15X?
Marvell Technology MRVL is currently trading at a discounted valuation, with its forward 12-month price-to-earnings (P/E) ratio at 7.15X, which is lower than the Zacks Electronics - Semiconductors industry average of 8.63X. Given MRVL's discounted valuation, investors might be wondering: Is this an opportunity to buy, or are there deeper challenges that could keep the stock in check? MRVL Forward 12 Months (P/E) Valuation Chart Product Innovation Positions MRVL for Sustainable Growth What makes Marvell Technology's low P/E value more attractive is its robust business prospects. Due to the proliferation of AI and high performance computing among data centers and hyperscalers, MRVL is experiencing massive traction in custom Application Specific Integrated Circuits, Custom high bandwidth memory Compute Architecture, Co-Packaged Optics Platform and Multi-Die Packaging Platform products. As data centers perform a growing number of AI-related tasks, improvements in networking, interconnect, processing and storage capabilities, requiring high-performance semiconductor solutions become crucial. Marvell Technology is capitalizing on this opportunity with 800G PAM, 400ZR DCI, and 1.6T PAM digital signal processing products. This growth is evident in Marvell Technology's data center segment, which has taken the lead among all its segments with 76% year-over-year revenue growth in the first quarter of fiscal 2026. Marvell Technology also plans to expand its customer base among hyperscaler customers that seek to stand out, cut expenses and want to gain more control over their AI infrastructure. Marvell Technology has collaborated with NVIDIA and leveraged the latter's NVLink Fusion platform to build comprehensive rack-scale AI solutions to meet the needs of hyperscalers. Furthermore, the shift from copper to optical connectivity in AI infrastructure represents a massive growth opportunity for Marvell Technology's Co-Packaged Optics technology. Marvell Technology is also experiencing a recovery among its enterprise networking and carrier infrastructure segments on the back of the demand rebound. However, Marvell Technology is also facing some challenges. With all these factors at play, the Zacks Consensus Estimate for Marvell Technology's 2026 revenues is pegged at $8.2 billion, indicating year-over-year growth of 42.6%. The consensus mark for earnings is pegged at $2.79 per share, suggesting a whopping 77.7% year-over-year increase. Key Challenges Faced by Marvell Technology Marvell is experiencing traction in its AI-focused custom silicon semiconductor business, but the margin in this business is half the story, as the margin in this business is fundamentally lower, affecting MRVL's gross margin. MRVL's custom AI silicon, including XPUs, is lowering MRVL's gross margins due to higher costs associated with manufacturing these chips. The ongoing macroeconomic uncertainties, like the U.S. government's evolving stance toward China, from which MRVL gained about 43% of its fiscal 2025 total revenues, are also a concern for the company. Investors' skepticism has also been intensified by the fear of sanctions and persistent tariff threats to China, where Marvell Technology owns research and development facilities. Furthermore, softness in MRVL's consumer end market due to volatility in gaming demand and lumpy order patterns in the industrial business has added to investor concerns. Marvell Technology also faces intense competition from Broadcom AVGO and Advanced Micro Devices AMD in the AI accelerator space and Micron Technology MU in the HBM space. Advanced Micro Devices is a strong player in the custom silicon solutions and AI accelerator space with its semi-custom SoC offerings and Instinct Accelerators that power numerous data centers. Advanced Micro Devices' reconfigurable Alveo Adaptable Accelerator Cards are used to speed up compute-intensive applications in data centers. Broadcom's advanced 3.5D XDSiP packaging platform is specifically designed to enhance the performance and efficiency of custom AI XPUs for AI accelerators. Micron Technology is also riding a powerful wave of demand for high-bandwidth memory (HBM) and DRAM products, especially as AI workloads surge. Micron Technology has made significant strides in AI-optimized memory solutions, with its HBM3E products gaining attention for their superior power efficiency and bandwidth. These factors have weighed on MRVL stock's performance. Stock Price Performance of MRVL Marvell Technology has underperformed the Zacks Electronics - Semiconductors industry in the year-to-date period by losing 32.6%. Marvell Technology YTD Performance Chart Conclusion: Hold MRVL Stock for Now Marvell Technology is facing several headwinds, including geopolitical tension, shrunken margins and growing competition across its end markets. However, the company has strong long-term fundamentals supported by its strong foothold in the data center and high-speed networking market. Considering all these factors, we suggest that investors should retain this Zacks Rank #3 (Hold) stock at present. You can see the complete list of today's Zacks #1 Rank (Strong Buy) stocks here. Zacks Names #1 Semiconductor Stock This under-the-radar company specializes in semiconductor products that titans like NVIDIA don't build. It's uniquely positioned to take advantage of the next growth stage of this market. And it's just beginning to enter the spotlight, which is exactly where you want to be. With strong earnings growth and an expanding customer base, it's positioned to feed the rampant demand for Artificial Intelligence, Machine Learning, and Internet of Things. Global semiconductor manufacturing is projected to explode from $452 billion in 2021 to $971 billion by 2028. See This Stock Now for Free >> Want the latest recommendations from Zacks Investment Research? Today, you can download 7 Best Stocks for the Next 30 Days. Click to get this free report Advanced Micro Devices, Inc. (AMD): Free Stock Analysis Report Micron Technology, Inc. (MU): Free Stock Analysis Report Marvell Technology, Inc. (MRVL): Free Stock Analysis Report Broadcom Inc. (AVGO): Free Stock Analysis Report
3 hours ago
Canadian teachers yearn for guidance, instruction as AI infiltrates classrooms
Even during the summer break, Ontario high school teachers Jamie Mitchell and Tamara Phillips know that many conversations they'll be having with their colleagues come fall will focus on the use of artificial intelligence in the classroom. Mitchell, who teaches math, and Phillips, who teaches English, are also instructional leaders, meaning they advise and lead their colleague on professional development. Mitchell says teachers are 100 per cent yearning for guidance on how to use AI in the classroom. By and large, there is a group of educators that feel that they need support in how to manage all the new issues that are coming up with respect to AI, he said. Enlarge image (new window) Ontario high school teachers Jamie Mitchell, left, and Tamara Phillips say their colleagues have received little guidance on how to incorporate AI into their instruction and lesson plans. Photo: CBC / Mark Bochsler Since ChatGPT, the chatbot from tech company OpenAI, first reared its head in student work almost two years ago, the increasing prominence of AI in everything from research to office work means conversations have turned from whether or not to allow it in the classroom, to how best to use it for educational purposes. Binary thinking around 'it's good or bad' should be tempered with the idea that, you know, learning and gaining knowledge about the tools that are at hand is really important, said Phillips. WATCH | Teachers seek AI instruction: Teachers are 'struggling on their own' Earlier this summer, tech companies Microsoft, OpenAI and Anthropic announced they would be investing $23 million US into the National Academy of AI Instruction — an initiative developed in partnership with the second largest teachers' union (new window) in the United States. The American Federation of Teachers says that over five years, the program aims to support 400,000 educators. According to the Canadian Teachers' Federation, the biggest teachers' union in Canada, educators here don't have anything like that, and certainly don't have anything that's been uniformly available across the country. Teachers are really struggling with artificial struggling on their own because there are no policies and frameworks put in place, said Heidi Yetman, who spoke to CBC from Ottawa during her time as the President of the Canadian Teachers' Federation. Her tenure ended last month. Enlarge image (new window) Heidi Yetman, the outgoing president of the Canadian Teachers' Federation, says the current guidance for teachers on AI isn't specific enough and doesn't offer true, meaningful education on the potential and pitfalls of AI in the classroom. Photo: CBC In Canada, places like Alberta and Quebec have rolled out AI guidance for schools, mostly focusing on what constitutes acceptable and unacceptable use. Yetman says the guidelines are a mishmash and all over the place. She says the guidance isn't specific enough and doesn't offer true, meaningful education for teachers on potential and pitfalls of AI in the classroom. CBC News reached out to provinces and territories to inquire about the level of support offered to teachers with respect to AI, and received a patchwork of responses. British Columbia said that while the province has provided general guidelines, school districts are developing their own local policies and approaches. The Newfoundland and Labrador Department of Education and Early Childhood Development said that in addition to general guidelines, to date, nearly 2,000 NLSchools staff members have participated in AI professional learning sessions. It noted that more training will be added to respond to the rapid evolution of AI technologies. LISTEN | How AI is changing education (new window) The diminishing role of teachers Teachers also may need reassurance that they'll still have a role in the brave new world where AI is part of education. In other words, can we use AI to replace a teacher? And, I hate to say it, but I think that we are headed in that direction, said Yetman. Johanathan Woodworth, the associate professor of education at Mount Saint Vincent University in Halifax, says anxiety around AI can change how teachers perceive themselves. Enlarge image (new window) Johanathan Woodworth, an associate professor of education at Mount Saint Vincent University in Halifax, says anxiety around AI can change how teachers perceive themselves. Photo: CBC For example, a lot of teachers are thinking, 'If I integrate AI, am I actually the teacher who owns this? Who is pedagogically in charge of the teaching?' Woodworth, who specializes in training aspiring teachers how to integrate technology into their teaching, says that no matter what form AI education for teachers takes in the future, teachers must be consulted. Who should train teachers on AI? One thing teachers, unions and professors of education seem to agree on is who shouldn't be paying for teacher training. Most expressed concern about the motivations of tech giants in the U.S. that are investing in teacher training, and wouldn't want to see it done in Canada. When the creators of various AI platforms are rolling out the learning, we aren't necessarily embedding what we talk about as humanized pedagogy and the practices that are needed in classrooms to build integrity and AI literacy into that learning, said Phillips, one of the teachers in Ontario. WATCH | How common is AI in student work? She and her colleague Mitchell say teachers are finding their own ways to use AI ethically. Mitchell was an early adopter of ChatGPT, working it into his lessons. One of the ways that math teachers have been using AI is to teach students how to turn, say, ChatGPT into a tutor with some very intelligent prompting. And then arming that student with the ability to ask ChatGPT for math help when they're at home, when they don't have access to their math teacher, he said. Mitchell says it's too late to pretend that AI won't infiltrate every aspect of education, including how students learn and complete their assignments, and how teachers assess them. The reality is today's AI is the worst AI that students are ever going to use, he said. And if teachers aren't arming those students with the skills to work with AI ethically, with integrity, then we're doing something wrong. With files from Griffin Jaeger
