Google DeepMind claims 'historic' AI breakthrough in problem solving
6 hours ago
- #AGI
- #Programming Competition
- #Artificial Intelligence
- Google DeepMind's Gemini 2.5 AI model achieved a 'historic' breakthrough by solving a complex real-world problem, winning a gold medal at an international programming competition.
- The AI outperformed top human teams from universities in Russia, China, and Japan, ranking second out of 139 participants.
- Google described the achievement as a 'profound leap in abstract problem-solving,' marking progress toward artificial general intelligence (AGI).
- The model was specially trained for coding, math, and reasoning, performing 'as well as a top 20 coder in the world.'
- Experts had mixed reactions: some praised the achievement, while others questioned its real-world impact and computing power requirements.
- Four key AI breakthroughs highlighted: Perceptron (1957), Deep Blue (1997), AlphaGo (2016), and AlphaFold (2020).