14 hours ago
ChatGPT took on a 50-year-old Atari — and lost
ChatGPT, arguably the world's most popular artificial intelligence (AI) chatbot, was humbled by a 50-year-old chess console. The
Atari 2600
, first launched in 1977, outperformed
ChatGPT
in a recent experiment to the point that the AI chatbot conceded.
The experiment was set up by cloud computing engineer Robert Jr. Caruso after ChatGPT itself insisted on playing chess against an Atari following a conversation about AI in chess.
OpenAI
's chatbot "claimed it was a strong player in its own right and would easily beat Atari's Video Chess", Caruso said in a LinkedIn post. He expected it to be "a lighthearted stroll down retro memory lane" for the AI. His hopes were deflated once the match began.
by Taboola
by Taboola
Sponsored Links
Sponsored Links
Promoted Links
Promoted Links
You May Like
Roteirizador Pathfind - O planejador de rotas mais completo do mercado
Sistema TMS embarcador
Saiba Mais
Undo
ChatGPT and its various AI models are being used for everything from answering emails, generating images and conducting in-depth research. Meanwhile, Atari 2600 sports an 8-bit, 1.1 MHz CPU which could only think 1–2 moves ahead.
Live Events
ChatGPT got absolutely wrecked on the beginner level, Caruso wrote. Despite being given a baseline board layout to identify pieces, ChatGPT confused rooks for bishops, missed pawn forks, and repeatedly lost track of pieces. The AI chatbot first blamed the Atari icons as too abstract to recognise, then failed to improve even after switching to the standard system of recording chess moves.
Discover the stories of your interest
Blockchain
5 Stories
Cyber-safety
7 Stories
Fintech
9 Stories
E-comm
9 Stories
ML
8 Stories
Edtech
6 Stories
At times, ChatGPT worked — analysing moves, explaining options and offering solid advice. At others, it made absurd suggestions — like sacrificing a knight to a pawn — or tried to move pieces that had already been captured, even during turns when it otherwise had an accurate view of the board, Caruso wrote.
He said he had to stop ChatGPT from making awful moves and correct its board awareness multiple times per turn over 90 minutes. The modern tech kept promising it would improve 'if we just started over.' Eventually, it conceded the challenge to Atari.
Technology has been better at chess than humans since the IBM Deep Blue supercomputer defeated the erstwhile champion and renowned grandmaster Gary Kasparov in 1997. The latest Stockfish
chess engine
have an estimated 3,600 points on the Elo chess ranking system, whereas the current world number one Magnus Carlsen has the highest rating among humans at around 2,800.
One response to Caruso's post argued that ChatGPT does not employ artificial general intelligence (AGI), and only emulates human examples. Thus, it was unfair to test it in a game of logic like chess. Another pointed out that ChatGPT is not a chess engine, unlike Atari 2600, which despite being archaic, tracks the board and formulates moves.
