August 12, 2026
King me, grandma of AI
Waits: Arthur Samuel's Checkers
The old-school computer checkers saga has readers geeking out over AI before AI was cool
TLDR: Arthur Samuel spent decades teaching early computers to play checkers, helping shape the first wave of artificial intelligence research. Readers were most struck by the twist that checkers—not just chess or Go—was a surprisingly big part of computing history.
Long before today’s chatbot chaos, Arthur Samuel was teaching a giant room-sized computer to play checkers—and the comment section is absolutely eating it up. The big mood? A mix of awe, nostalgia, and a little disbelief that one of the earliest machine-learning stories wasn’t about chess at all. The standout reaction came from one reader who basically summed up the thread’s vibe: people thought early computer brains obsessed over chess and Go, so finding out checkers was secretly a major star of the show felt like a delightful plot twist.
That’s the heart of the story here. Samuel started tinkering with the idea in the 1940s, got it running at IBM in the 1950s, showed it off on television, and eventually built a program that could give strong human players a real challenge. Later versions even ran on old university systems and a tiny home game console with just 1KB of memory, which today sounds almost like making a gourmet meal with a toaster and a paper clip. Readers seemed charmed by that sheer stubbornness: this wasn’t flashy modern AI hype, but a decades-long obsession powered by patience, awkward hardware, and pure nerd determination.
There isn’t huge comment-war drama yet, but there is a clear hot take brewing: maybe the history of AI is way weirder, broader, and more playful than people assume. And honestly, the community seems delighted by that.
Key Points
- •Arthur Samuel began planning a checkers-playing computer system at the University of Illinois before moving to IBM in 1949, where he continued the project.
- •Samuel said a first version of the program was ready in 1952, and by 1954 on the IBM 704 it could play an interesting game with added learning capability.
- •The program was demonstrated on television in 1956 and later reached a level that could sometimes challenge a master player.
- •The article identifies minimax, alpha-beta pruning, and rote learning as techniques used in Samuel’s influential early AI work.
- •A later version of the program ran on WAITS after being ported at SAIL, and the article explains how users can still run it and inspect board states and search statistics.