AI’s 48-Hour Frenzy: Breakthroughs, Billion-Dollar Bets & Bold Policies (July 22–23, 2025)
AI Tackles Advanced Math: In a milestone for AI reasoning, models from Google and OpenAI achieved gold-medal scores at the International Mathematical Olympiad reuters.com. It marks the first time AI systems crossed the Olympiad’s top-tier threshold, each solving 5 out of 6 challenging problems using new “general-purpose reasoning” methods that rely on natural language understanding reuters.com reuters.com. “I think the moment we can solve hard reasoning problems in natural language will enable the potential for collaboration between AI and mathematicians,” said Brown University professor Junehyuk Jung, suggesting AI may soon help tackle unsolved research problems in math reuters.com reuters.com. OpenAI’s team disclosed that their experimental model scaled up “test-time compute,” allowing the AI to “think” longer and in parallel – a costly effort the company called “very expensive” reuters.com. Google DeepMind, which worked directly with IMO organizers, used a general model Gemini Deep Think to solve problems within the official 4.5-hour limit entirely in natural language reuters.com. Experts at both companies see this as a breakthrough hinting that AI’s reasoning power could soon extend to frontiers in physics and other sciences reuters.com.