5 Key Takeaways
- Magnus Carlsen humorously commentated as Elon Musk’s Grok 4 was defeated 4-0 by OpenAI’s o3 in an AI chess tournament.
- Grok 4 made several major blunders in the final, including sacrificing key pieces and losing its queen multiple times.
- The tournament featured eight large language models (LLMs), with o3 taking first place and Google’s Gemini 2.5 Pro finishing third.
- Carlsen rated Grok’s chess strength at 800 and o3 at 1200, noting that LLMs are still far behind specialized chess engines.
- Carlsen criticized the overall chess abilities of the LLMs, saying most played like beginners and made inexplicable moves.
Magnus Carlsen Has a Laugh as Elon Musk’s Grok 4 Gets Crushed by OpenAI’s o3 in AI Chess Showdown
If you thought artificial intelligence was ready to take over the chess world, think again! In a recent online chess tournament featuring some of the world’s most advanced AI chatbots, things didn’t go quite as planned for Elon Musk’s Grok 4. The event, held on Google’s Kaggle Game Arena, pitted eight large language models (LLMs) against each other—including Google’s Gemini, Anthropic’s Claude, and two from OpenAI. But the real drama unfolded in the final, where Grok 4 faced off against OpenAI’s o3.
Five-time world chess champion Magnus Carlsen was on hand to provide live commentary—and he didn’t hold back. As Grok 4 made one baffling mistake after another, Carlsen and fellow grandmaster David Howell couldn’t help but laugh and shake their heads in disbelief.
Grok 4 had looked strong in earlier rounds, but in the final, it fell apart. In the very first game, Grok gave away its bishop on move 8 for no good reason, then started trading off its other pieces—even its queen, the most powerful piece on the board! Carlsen compared it to “watching kids’ games,” and joked that everyone should feel better about their own chess after seeing this.
The match was a clean sweep: OpenAI’s o3 won all four games. Carlsen rated Grok’s chess skills at about 800 (beginner level), while o3 scored a more respectable 1200 (club player level). For context, top human players are rated over 2500.
Carlsen’s commentary was full of zingers. When Grok blundered its queen in game two, he said it was like “that one guy in a club tournament who knows the opening but nothing else.” In game three, after another queen blunder, Carlsen burst out laughing, saying Grok “thinks it’s playing giveaway or something.”
The tournament also had a bit of tech world drama. OpenAI’s Sam Altman and Elon Musk, once co-founders, are now rivals. Musk even sued OpenAI last year, claiming it broke its promise to put public good over profit.
So, what’s the takeaway? While AI engines like AlphaZero and Leela are already stronger than any human, these chatbots still have a lot to learn about chess. For now, it seems, even the world’s best AI chatbots can make mistakes that would make a beginner blush—and give Magnus Carlsen a good laugh in the process.
No comments:
Post a Comment