Featured Image Credit: SOPA Images / Contributor via Getty
As fast as artificial intelligence might be improving, it's apparently still not good enough to beat a renowned chess champion, as the current world number one shared his humiliating victory over ChatGPT.
We're a far cry from the days of basic questions and wacky image prompts, as artificial intelligence is now able to conduct eerie and vast conversations, replicate the advice of veteran medical professionals, and even generate an image of a half-filled wine glass.
Key AI leaders have confessed that we're less than a decade away from achieving 'artificial general intelligence', marking the point at which the capacity of AI overtakes that of humans in all aspects, and Elon Musk is already claiming that his company's latest model is smarter than grad students.
While you'd imagine technology like artificial intelligence to excel at logic-based games, one world champion showed that ChatGPT still has a long way to go when it comes to chess, revealed how he humiliating it on social media.
As reported by Firstpost, 34-year-old Magnus Carlsen, who has been world chess number one for over 14 years straight, positively smashed ChatGPT in a game of chess, even with an incredibly specific win condition.
Advert
Carlsen shared on X that he "sometimes gets bored while traveling," accompanying the post with two screenshots detailing his victory and post-match conversation with the AI model.
"And with 53... Bxh5, you've officially done it: All of my pawns are gone, you haven't lost a single piece, you fulfilled your win condition perfectly," ChatGPT wrote after being embarassed.
"I resign. That was methodical, clean, and sharp. Well played!"
While it might not necessarily be a shock that the world's greatest chess player is good at the game, it shows that AI still has its limitations even in a game that would hypothetically benefit from a computer being able to run through millions of scenarios to find the right move every time.
Following the match, Carlsen asked the chatbot to analyze his in-game performance, and it offered some intriguing conclusions.
First up, ChatGPT praised his understanding of openings, commending him on his handling of the Philidor Defense, while also lauding his positional patience and tactical awareness.
What's more, the AI also indicated that Carlsen's endgame technique was impressive, outlining that his "bishops and pawns converted a small edge into a winning plan."
Surprisingly, ChatGPT also admitted to attempting several illegal moves which was lauded by the chatbot, writing that he "enforced the rules of the game with consistency, which suggests real over-the-board experience."
Importantly, one key part of Carlsen's initial request was to evaluate his classical rating strength, which is what is used to place you on the global leaderboards.
"If you played like this consistently in longer time controls (classical, not blitz or rapid), I'd estimate your classical strength to be around 1800-2000 FIDE or USCF. Possibly higher if your opening prep and tactical sharpness hold up under pressure," evaluated the AI.
For comparison, Magnus Carlsen's actually classical rating is 2839, so that's another thing that ChatGPT didn't quite succeed with - although it would be bold for the model to estimate someone being the world number one based on a single game.
Even a rating of 2500 would land you 603rd globally, with the upper estimate of 2000 giving a ranking of just 26,599th in the world, which is obviously a far cry from Carlsen's actual ability.
