Technology. The ever-advancing pinnacle of the processor. It’s a modern-day blessing to most, but dreaded by those convinced it’s five minutes away from flipping society onto its head like an egg served sunny-side down.
The word alone can strike fear into the hearts of man. Especially if you shout it loud enough into their ears as you pass them on the street.
And, if you were to take a small break from harassing the general public and ask one of them to name today’s most advanced piece of technology, they’d likely say ChatGPT — OpenAI’s super-brainy chatbot, packed with enough artificial intelligence to seemingly make regular intelligence look like its eating glue from a pot in a sandbox.
But if ChatGPT, the poster child for cutting-edge technology, is so smart, how did it just get absolutely bodied by a video game console released in 1977?
ChatGPT vs. Atari 2600: All hail the golden oldie?
When we think of competitors to ChatGPT, there’s a usual list of suspects to choose from: Google Gemini, Microsoft Copilot, Meta AI, Anthropic’s Claude, Perplexity AI, and maybe even newcomer DeepSeek.
What you wouldn’t expect to appear in that list is the Atari 2600, a 48-year-old home video game console best known for bringing Pac-Man into the living rooms of millions of first-generation gamers.
However, thanks to Citrix Engineer Robert Jr. Caruso, Atari’s retro console can now be counted among ChatGPT’s truest rivals, after it was used to repeatedly best OpenAI’s GPT-4o model at a simple game of chess.
In a now-viral post shared to LinkedIn, Caruso details his 90-minute experiment in pitting the computing might of tens of thousands of Nvidia GPUs against the singular 1.19 MHz 8-bit MOS Technology 6507 processor of the Atari 2600, claiming ChatGPT “made enough blunders to get laughed out of a 3rd grade chess club.”
A short lived victory for retro tech
Clearly, they don’t make them like they used to. The prevalence of the 2600 over ChatGPT is a true David vs. Goliath battle on a checkerboard stage, and a surprising outcome to most.
However, there’s a chasm of difference between chatbot and a chess engine, meaning the Atari 2600, which ran not as hardware but through the Stella emulator, likely had ChatGPT’s number from the start. While OpenAI has made great strides in improving its model’s memory capabilities, it’s still primarily a language prediction machine, and not the next Deep Blue.
Still, given the 2600’s ability to only predict two moves in advance, it does highlight ChatGPT’s shortcomings, and provide a thumb to the virtual eye for OpenAI’s world’s most intelligent chatbot.
At least it would do if OpenAI had ever claimed as much. In fact, it’s more often than not proclaiming the opposite.
Stretching back to 2023, in an episode of the Lex Friedman podcast, OpenAI CEO Sam Altman was quick to label GPT-4 as “a very early AI. It’s slow, it is buggy, and it does not do a lot of things very well.” Then, in 2024, during a Q&A at Stanford University, Altman claimed that ChatGPT was running on “the dumbest model any of you will ever have to use again by a lot.”
In fact, only recently has Altman pushed the message that AI is living up to its supposed smarts. In a blog post published on Tuesday, titled The Gentle Singularity, Altman predicts: “We do not know how far beyond human-level intelligence we can go, but we are about to find out.”
So yes, Atari’s classic console may have bested ChatGPT this time, but if Altman’s words are anything to go by, it could be a very short-lived victory.
