AI models win gold at global maths contest
But neither model scored full marks — unlike five young humans at the International Mathematical Olympiad (IMO), a prestigious annual competition where participants must be under 20 years old.
The results marked the first time that AI systems crossed the gold-medal scoring threshold at the contest held annually for high-school students.
Five Thais competed in the event that concluded on Saturday in Queensland, Australia, with one entrant — Patanasang Pinijpichitkul — winning gold with a score of 95.87%.
Both the Google and OpenAI models solved five out of the 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.
While Google DeepMind worked with the IMO to have its models graded and certified, OpenAI did not officially enter the competition. The latter said its models achieved a gold medal-worthy score, citing grades by three external IMO medalists.
Unlike previous AI attempts that relied on formal languages and lengthy computation, Google said its approach this year operated entirely in natural language and solved the problems within the official 4.5-hour time limit.
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 professor at Brown University and visiting researcher at the 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 said its breakthrough was achieved with a new experimental model centred 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, a 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.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Bangkok Post
9 hours ago
- Bangkok Post
GFCS Unveils New Brand to Drive Smart Security
Bangkok, July 28, 2025 — Guardforce Cash Solutions Security (Thailand) Co., Ltd. (GFCS), a leader in secure logistics and asset protection, has unveiled a refreshed brand identity that marks a strategic pivot towards technology-driven innovation. The rebranding, showcased during the company's high-profile 'Beyond Cash' event in Bangkok, signifies GFCS's transformation from a traditional cash-handling enterprise into a forward-thinking provider of AI-powered business solutions. The new brand design embraces the symbolic shield emblem of the Guardforce AI Group, embodying reliability, innovation, and collaboration. More than visual—the shift reflects GFCS's expanding capabilities across smart security technologies, automation, and integrated systems that meet the needs of a fast-changing economy. No longer limited to conventional cash logistics, the company now supports a wide spectrum of asset management services—from jewellery and electronics to sensitive corporate documents. Olivia Wang, Chairwoman and CEO of Guardforce AI, emphasised the significance of this transformation, stating that it merges over 40 years of operational expertise with cutting-edge artificial intelligence, robotics, and biometric authentication technologies. She noted that the goal is to create smarter, safer, and more adaptable services that empower clients to operate with enhanced speed and precision. 'Today's rebranding is not just about a logo change; it's about embracing transformation while honouring our legacy. We are transforming from manpower-base security company into a high-tech solutions company'. Central to GFCS's new offering are its Smart Cash and Smart Retail solutions, which embody the company's flagship innovations. Smart Cash provides financial institutions and retailers with advanced capabilities such as predictive cash flow analytics and automated reconciliation, optimising liquidity and security. While Smart Retail integrates AI and IoT technologies to offer real-time asset tracking and performance analytics, streamlining operations and enabling better customer engagement. These innovations serve as the foundation for GFCS's three strategic transformations: 'Beyond Just Cash,' 'Beyond Logistics,' and 'Beyond Services.' The company now manages not only currency but also high-value assets like jewellery, electronics, and sensitive documents. It's shifting from basic transport to smart, closed-loop systems equipped with predictive analytics, smart safes, and real-time monitoring. Through AI, IoT, and biometrics, GFCS is evolving into a full-scale solution provider—developing intelligent infrastructure in partnership with Guardforce AI's global R&D. Kelvin Chu, Managing Director of GFCS Thailand, reiterated that while the company's legacy rests on reliability and trust, its future will be defined by agility, intelligence, and growth. 'GFCS's transformation into 'Beyond Cash' reflects our dedication to evolving alongside our clients. This is not a departure from our roots but an elevation of our service to the new heights.' The 'Beyond Cash' event gathered top executives, technology partners, and media to commemorate this turning point and explore the company's enhanced capabilities. The event concluded with a shared vision: to reshape asset management in Thailand using innovation as the catalyst. GFCS's rebranding positions it at the forefront of the secure services industry, ready to meet the demands of a rapidly digitalizing market; showing confidence and commitment in future business development.
Bangkok Post
22-07-2025
- Bangkok Post
AI models win gold at global maths contest
Google and OpenAI said their artificial-intelligence models won gold medals at a global mathematics competition, signalling a breakthrough in capabilities in the race to build systems that can rival human intelligence. But neither model scored full marks — unlike five young humans at the International Mathematical Olympiad (IMO), a prestigious annual competition where participants must be under 20 years old. The results marked the first time that AI systems crossed the gold-medal scoring threshold at the contest held annually for high-school students. Five Thais competed in the event that concluded on Saturday in Queensland, Australia, with one entrant — Patanasang Pinijpichitkul — winning gold with a score of 95.87%. Both the Google and OpenAI models solved five out of the 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. While Google DeepMind worked with the IMO to have its models graded and certified, OpenAI did not officially enter the competition. The latter said its models achieved a gold medal-worthy score, citing grades by three external IMO medalists. Unlike previous AI attempts that relied on formal languages and lengthy computation, Google said its approach this year operated entirely in natural language and solved the problems within the official 4.5-hour time limit. 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 professor at Brown University and visiting researcher at the 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 said its breakthrough was achieved with a new experimental model centred 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, a 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.
Bangkok Post
22-07-2025
- Bangkok Post
Artificial intelligence boosts battle against illegal websites
Artificial intelligence has been deployed, enabling the shutdown of 100,000 gambling and porn sites per month, although more needs to be done to curb these illegal sites, according to the Ministry of Digital Economy and Society (DES). The DES now blocks up to 100,000 illegal websites per month, a tenfold increase from the previous monthly rate of 10,000, DES Minister Prasert Jantararuangtong said. However, challenges remain due to the adaptability of criminal networks and the foreign-based infrastructure of many of these platforms. The update came during a Senate session in parliament yesterday, in response to an inquiry by Senator Dr Premsak Piayura, who raised concerns about ongoing cybercrime -- especially call centre scams, online gambling, and explicit content websites. Dr Premsak questioned why, despite the recent enforcement of the executive decree on the prevention and suppression of technological crime, citizens continue to fall victim to online fraud. "Scammers today are more cunning than ever. Even with repeated warnings, many realise the danger only after they've lost their money," he said. He pointed specifically to mobile operators, who now face increased legal responsibilities under the new law. Dr Premsak also inquired whether the newly established Cyber Crime Suppression Center (CCSC) was effectively functioning and whether inter-agency coordination remained an obstacle, a known weakness in Thai bureaucracy. Mr Prasert assured senators that coordination among agencies has improved, noting that since the decree came into force in May, the government has implemented financial, digital asset, and telecommunications measures, as well as created mechanisms for the reimbursement of victims. He said the CCSC had now been legally formalised and explained its central data platform has strengthened law enforcement's ability to act, significantly reducing the overall damage caused by scams. On the "mule accounts", the minister acknowledged improvements in tracing and shutting them down. However, the more stubborn problem remains the proliferation of gambling and pornographic websites. "In the past, we could only shut down about 10,000 sites a month. But today, with AI developed by the DES Ministry, we're able to block around 100,000 sites monthly," Mr Prasert said. "But the problem hasn't gone away -- these sites continuously change URLs and operate on overseas servers, making it hard to identify the operators and prosecute them." He added that while technical capacity has grown, legal challenges persist, particularly in identifying and taking action against foreign-registered platforms. Dr Premsak, meanwhile, urged the ministry to find ways to pressure foreign platforms, such as Facebook, to take responsibility when their services are used to defraud Thais. "We need to know how to bring these platforms under Thai jurisdiction or involve them in restitution when damage occurs," he emphasised. Mr Prasert reiterated that while technical solutions like AI have increased the government's ability to block harmful content, legal and diplomatic strategies will be needed to tackle foreign-based digital crime infrastructures. He also confirmed that the ministry plans to expand its outreach to vulnerable groups to help them avoid traps and know how to respond if they fall victim.
