Google DeepMind has claimed that its new Gemini 2.5 AI model has achieved a historic milestone. The company compared this moment to when IBM’s Deep Blue defeated chess world champion Garry Kasparov in 1997, or when AlphaGo beat a human Go champion in 2016. The difference, however, is that this time the AI solved not a game but a real-world complex problem.
Google DeepMind’s Historic AI Breakthrough: Gemini 2.5 Solves Complex Problem
At an international programming competition in Azerbaijan, the AI was challenged to figure out how to distribute liquid through a network of pipes and reservoirs in the fastest and most efficient way possible. The problem was extremely tough because it required analyzing countless possibilities and choosing the right path. Out of 139 top human programmer teams, not a single one could solve it, but the AI cracked the solution in less than 30 minutes. Out of 12 tasks, it succeeded in 10 and secured the second overall rank. According to Google, its performance was equivalent to that of a top 20 programmer in the world.
Google DeepMind described this as a historic step toward AGI (Artificial General Intelligence). They believe this achievement proves that AI can now reason not only in games or limited scenarios but also in real-world challenges. The company added that such capabilities could in the future revolutionize fields like drug discovery, chip design, and engineering.
However, not all experts fully agree with Google’s claims. Stuart Russell, a professor at UC Berkeley, said that the “claims of historic importance” might be exaggerated. According to him, even Deep Blue’s victory in chess had little impact on real-world AI development. Still, he acknowledged that AI solving a programming contest task correctly shows progress toward making AI systems more reliable and capable of producing high-quality code. Similarly, Michael Wooldridge, professor at Oxford University, called it an “impressive achievement” but raised the question of how much computing power was required for this feat.
Google admitted that the model used more computational resources than what’s available in its standard $250 per month AI Ultra service, but did not disclose exact figures. Meanwhile, Dr. Bill Poucher, executive director of ICPC, said that Gemini’s success at this level marks an important moment in defining AI tools and academic standards for the next generation.
In conclusion, this achievement demonstrates that AI is no longer limited to winning games but is now capable of tackling real-world complex problems. While experts caution that it is still too early to call this the arrival of AGI, there is little doubt that this breakthrough could prove to be a milestone in the evolution of artificial intelligence.
Also Read:
What is Artificial Intelligence? A Complete Beginner’s Guide with Examples