Humans beat AI gold-level score at top maths contest
Neither model scored full marks — unlike five young people at the International Mathematical Olympiad (IMO), a prestigious annual competition where participants must be under 20 years old.
Google said Monday that an advanced version of its Gemini chatbot had solved five out of the six maths problems set at the IMO, held in Australia's Queensland this month.
'We can confirm that Google DeepMind has reached the much-desired milestone, earning 35 out of a possible 42 points — a gold medal score,' the US tech giant cited IMO president Gregor Dolinar as saying.
'Their solutions were astonishing in many respects. IMO graders found them to be clear, precise and most of them easy to follow.'
Around 10 percent of human contestants won gold-level medals, and five received perfect scores of 42 points.
US ChatGPT maker OpenAI said that its experimental reasoning model had scored a gold-level 35 points on the test.
The result 'achieved a longstanding grand challenge in AI' at 'the world's most prestigious math competition,' OpenAI researcher Alexander Wei wrote on social media.
'We evaluated our models on the 2025 IMO problems under the same rules as human contestants,' he said.
'For each problem, three former IMO medalists independently graded the model's submitted proof.'
Google achieved a silver-medal score at last year's IMO in the British city of Bath, solving four of the six problems.
That took two to three days of computation — far longer than this year, when its Gemini model solved the problems within the 4.5-hour time limit, it said.
The IMO said tech companies had 'privately tested closed-source AI models on this year's problems,' the same ones faced by 641 competing students from 112 countries.
'It is very exciting to see progress in the mathematical capabilities of AI models,' said IMO president Dolinar.
Contest organizers could not verify how much computing power had been used by the AI models or whether there had been human involvement, he cautioned.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Arab News
3 days ago
- Arab News
How AI speech-to-text technology is tuning in to a digital Saudi Arabia
DHAHRAN: In a world racing toward automation, Klemen Simonic believes the most natural interface is also the most enduring: the human voice. As founder and CEO of Soniox — a cutting-edge speech-to-text platform — Simonic is betting that voice-powered technology will drive the next wave of digital innovation. And in a country like Saudi Arabia, where smartphones dominate daily life and a young population is hungry for digital solutions, the potential is hard to ignore. Soniox, which Simonic launched five years ago, offers speech recognition, transcription and real-time multilingual translation in more than 60 languages. Unlike many competitors, it delivers ultra-fast, token-level outputs in milliseconds — a critical advantage for live assistants, wearables, bots and smart speakers. But Simonic's journey toward building the company began long before the rise of generative AI. 'I started in programming development right after high school, and I was invited to join the Jozef Stefan Institute in Slovenia, one of the best institutes in this part of Europe,' he told Arab News. 'I was working there with Ph.D. students and postdocs on machine learning, natural language processing, dependency parsing, tokenization, tagging and entity extraction.' That early exposure led him to two internships at Stanford University in 2009 and 2011, where he worked alongside top researchers in AI. 'I wanted to join Google to work on these cool things,' he said. After an internship there in 2014, Simonic was courted by both Google and Facebook — ultimately joining the latter in 2015 to help build speech recognition systems now used across Facebook, Instagram and WhatsApp. Today, his company is focused entirely on voice AI, and its promise goes beyond convenience. With privacy and compliance built in — including SOC 2 Type II certification and HIPAA readiness — Soniox is already being used in hospitals, call centers and emergency rooms where clear, accurate transcription can be a life-saving tool. 'We have many healthcare customers using our API in emergency rooms where real-time AI interpretation can bridge communication gaps that human translators sometimes cannot, especially with complex medical terminology,' said Simonic. Saudi Arabia represents a particularly compelling market for the company's ambitions. With more than 90 percent smartphone penetration and a population where 70 percent of people are aged under 35, the Kingdom is fertile ground for voice-enabled technologies. The widespread adoption of government-developed platforms like Tawakkalna during the COVID-19 pandemic only accelerated the Kingdom's reliance on mobile-first services. 'Data and artificial intelligence contribute to achieving Saudi Arabia's Vision 2030; this is because, out of 96, 66 of the direct and indirect goals of the vision are related to data and AI,' according to the Saudi Data & AI Authority. The Kingdom's communications and IT sector is now worth more than $44 billion — 4.1 percent of gross domestic product — and expanding quickly with strategic investments in cloud computing, automation and smart infrastructure. Although Soniox does not yet have a team on the ground in the region, the company sees significant interest from Saudi organizations exploring AI-powered transcription and customer service tools. Simonic said there are pilot programs in countries like Portugal and interest from companies in Saudi Arabia looking to improve call center and transcription services. And while Arabic remains one of the more complex languages for voice AI, Simonic sees both the challenge and the opportunity. Many of Saudi Arabia's rural communities speak dialects rich in cultural nuance — languages that are often excluded from mainstream datasets. This environment offers fertile ground for Soniox's technology, which strives to 'enable all languages, so everyone in the world can speak and be understood by AI.' Simonic's team, primarily based in Slovenia, is committed to expanding language support to make the technology more inclusive, even in markets where none of the developers speak the local tongue. Soniox is also designed with flexibility in mind. Businesses can integrate its API without storing any audio or transcripts, ensuring tight data control. For individual users, features like encrypted transcripts and a summarizing tool enhance productivity — even for the tech-averse. 'My mom is not very tech-savvy, but she uses our app to build her grocery shopping list,' Simonic said. 'That was not the original purpose, but it shows how technology can evolve in ways we didn't expect.' In July, Soniox launched a new comparison tool that allows developers and businesses to benchmark different speech AI providers using their own voice samples and real-world data. It is another step toward transparency and broader adoption — especially in regions like the Gulf, where choosing the right solution can hinge on performance in diverse linguistic contexts. 'The tech morphs, but the human voice remains the most intimate and effective way we communicate,' Simonic said. As Saudi Arabia pushes forward with its digital transformation under Vision 2030, technologies like Soniox may find their voice amplified — not just as a tool for productivity, but also as a bridge between language, innovation and access in a rapidly changing world.
Asharq Al-Awsat
3 days ago
- Asharq Al-Awsat
Google-parent Alphabet Earnings Shine with Help of AI
Google-parent Alphabet on Wednesday reported quarterly profits that topped expectations, saying artificial intelligence has boosted every part of its business. Alphabet's second-quarter profit of $28.2 billion -- on $96.4 billion in revenue -- came with word that the tech giant will spend $10 billion more than it previously planned this year on capital expenditures, as it invests to meet growing demand for cloud services. "We had a standout quarter, with robust growth across the company," said Alphabet chief executive Sundar Pichai. "AI is positively impacting every part of the business, driving strong momentum." Revenue from search grew double digits in the quarter, with features such as AI Overviews and the recently launched AI mode "performing well," according to Pichai. Ad revenue at YouTube continues to grow along with the video platform's subscription services, Alphabet reported. Alphabet's cloud computing business is on pace to bring in $50 billion over the course of the year, according to the company. "With this strong and growing demand for our cloud products and services, we are increasing our investment in capital expenditures in 2025 to approximately $85 billion and are excited by the opportunity ahead," Pichai said. Alphabet shares were up nearly 2 percent in after-market trades that followed the release of the earnings figures. Investors have been watching closely to see whether the tech giant may be pouring too much money into artificial intelligence and whether AI-generated summaries of search results will translate into fewer opportunities to serve up money-making ads. The internet giant is dabbling with ads in its new AI Mode for online search, a strategic move to fend off competition from ChatGPT while adapting its advertising business for an AI age. The integration of advertising has been a key question accompanying the rise of generative AI chatbots, which have largely avoided interrupting the user experience with marketing messages. However, advertising remains Google's financial bedrock. "Google is doing well despite tariff headwinds and rising AI competition in search," said eMarketer principal analyst Yory Wurmser. "It's also successfully monetizing AI Overviews and AI Mode, a good sign for the future." Google and rivals are spending billions of dollars on data centers and more for AI, while the rise of lower-cost model DeepSeek from China raises questions about how much needs to be spent. Antitrust battles Meanwhile the online ad business that generates the cash Google invests in its future could be neutered due to a defeat in a US antitrust case. During the summer of 2024, Google was found guilty of illegal practices to establish and maintain its monopoly in online search by a federal judge in Washington. The Justice Department is now demanding remedies that could transform the digital landscape: Google's divestiture from its Chrome browser and a ban on entering exclusivity agreements with smartphone manufacturers to install the search engine by default. District Judge Amit Mehta is considering "remedies" in a decision expected in the coming days or weeks. In another legal battle, a different US judge ruled this year that Google wielded monopoly power in the online ad technology market, another legal blow that could rattle the tech giant's revenue engine. District Court Judge Leonie Brinkema ruled that Google built an illegal monopoly over ad software and tools used by publishers. Combined, the courtroom defeats have the potential to leave Google split up and its influence curbed. Google said it is appealing both rulings.
Arab News
6 days ago
- Arab News
Humans beat AI gold-level score at top maths contest
SYDNEY: Humans beat generative AI models made by Google and OpenAI at a top international mathematics competition, despite the programs reaching gold-level scores for the first time. Neither model scored full marks — unlike five young people at the International Mathematical Olympiad (IMO), a prestigious annual competition where participants must be under 20 years old. Google said Monday that an advanced version of its Gemini chatbot had solved five out of the six maths problems set at the IMO, held in Australia's Queensland this month. 'We can confirm that Google DeepMind has reached the much-desired milestone, earning 35 out of a possible 42 points — a gold medal score,' the US tech giant cited IMO president Gregor Dolinar as saying. 'Their solutions were astonishing in many respects. IMO graders found them to be clear, precise and most of them easy to follow.' Around 10 percent of human contestants won gold-level medals, and five received perfect scores of 42 points. US ChatGPT maker OpenAI said that its experimental reasoning model had scored a gold-level 35 points on the test. The result 'achieved a longstanding grand challenge in AI' at 'the world's most prestigious math competition,' OpenAI researcher Alexander Wei wrote on social media. 'We evaluated our models on the 2025 IMO problems under the same rules as human contestants,' he said. 'For each problem, three former IMO medalists independently graded the model's submitted proof.' Google achieved a silver-medal score at last year's IMO in the British city of Bath, solving four of the six problems. That took two to three days of computation — far longer than this year, when its Gemini model solved the problems within the 4.5-hour time limit, it said. The IMO said tech companies had 'privately tested closed-source AI models on this year's problems,' the same ones faced by 641 competing students from 112 countries. 'It is very exciting to see progress in the mathematical capabilities of AI models,' said IMO president Dolinar. Contest organizers could not verify how much computing power had been used by the AI models or whether there had been human involvement, he cautioned.
