In a chess exhibition tournament on Google’s Kaggle Game Arena, Elon Musk’s AI model Grok 4 was decisively defeated 4-0 by OpenAI’s o3. The event featured eight large language models (LLMs), including competitors like Gemini 2.5 Pro and Claude 4 Opus. Despite initial expectations, Grok 4 struggled in the final, making uncharacteristic sacrifices and blunders, prompting laughter from Magnus Carlsen, who provided live commentary. Each game showcased Grok’s questionable decision-making, such as sacrificing key pieces without strategic benefit. Carlsen humorously evaluated Grok’s chess strength at 800, while o3 was estimated at 1200. Following Grok’s losses, Carlsen critiqued the overall performance of the AI models, noting that most, including Claude, fell short of expectations. The tournament underscored the fierce rivalry between Musk and OpenAI’s Sam Altman, stemming from their shared history as co-founders of OpenAI.
Source link
