AI and Games
AI and Games is the category of the moments when ordinary people met a computer that "understood" them, or a computer that played against them.
This English edition covers selected entries, chosen for translation from the full Hebrew site. More parts of the site are being translated gradually.
Questions this category answers
How did a home computer from the 1980s "understand" what I typed, if it had no artificial intelligence?
Why did AI researchers choose games like chess and Go to test their machines?
What is the difference between a game character that looks smart and one that actually learns?
How did we get from a two-word command to a game you can write any sentence to?
What changed when a computer beat the world chess champion?
Where to start
- Text adventures and the two-word parser If you read only oneA two-word parser is how early text adventure games turned a line of text the player typed into an action in the game: one verb and one noun, nothing more…
- Deep BlueDeep Blue was IBM's chess computer that beat world champion Garry Kasparov in 1997. It was the first time a computer defeated a reigning world champion in…
- AI DungeonAI Dungeon is a text adventure game by the company Latitude, in which a language model writes the story as you play. Its first versions were built on GPT-…
More about this category
Its story starts with GET LAMP and reaches ChatGPT. In the late 1970s, players typed two words into a text adventure, and the computer answered as if it understood. Behind this were rules written in advance, not learning, but it was one of the first encounters of a wide audience with a machine you talk to in human language.
Games were also the laboratory of AI researchers: checkers, chess and Go are closed worlds, with clear rules and a result you can measure, so they were where researchers tested again and again whether a machine can learn and plan. The category follows these two lines: the parsers of text adventures, game characters that act on their own, computers that beat the best players, and today's language models, which let you write a free sentence to a game and get an answer. It is new, and it will grow gradually.
What is not here
Here you will find real systems: a computer that plays, a game you talk to, and game characters that behave on their own. A fictional work that depicts a thinking machine, even when it is a computer game, is covered in AI in Popular Culture. The home computers the games ran on are in Digital Foundations, and the methods themselves, such as reinforcement learning, are in AI Foundations. Entries such as Deep Blue and AlphaGo belong to History and Key People, and are shown here too.
Z
📋 Criteria for entries in this category
An entry enters here if it is about a real system in which games and artificial intelligence meet, in one of four ways: a computer that plays against a person or learns to play; a game you talk to in human language, such as text adventures and their parsers; characters and worlds in a game that behave on their own; and a game that serves as a laboratory for research. The test: the entry has something to say about artificial intelligence, not only about the game. A game without such a story, a home computer or a game company is not included. Much of the "intelligence" in old games is rules written in advance, and the entries say so. A fictional work that depicts a thinking machine is in AI in Popular Culture. Entries from other categories, such as Deep Blue and AlphaGo, are shown here too and stay in their own category.