Timeline
Older AI relied on humans designing complicated but explicit instructions for how to interpret input and generate output. At their most basic, these instructions, or rules, take the form "if X, then Y." "If a traffic light is green, then cars can go." "If text includes the phrase 'Dear So-and-so', then it is a letter or email." "If someone listens to Lana Del Ray, then they may also like Billie Eilish."
Artificial Intelligence become a thing? When did the technology that underpins current AI models start being developed? There's no easy answer. As such, we've decided to give a general description of "how we got here" followed by some detailed timeline resources for your perusal depending on your needs.
We'll start with a story.
There was once a fancy chat bot named ELIZA. There are many ways for you to try ELIZA out yourself online, but here's a version hosted in a web browser by Cal State Fullerton you can try right now. ELIZA was a cool parlor trick that pulled proper nouns and other important words from your sentences based on how sentences work and then used that to prompt you to keep talking in an attempt to mimic Rogerian psychotherapy. It's creator thought this was an interesting exercise in how to program machines. However, many people who used it went, "Woah, this machine is now a real thinking person who is fully sentient and now computers will take over everything and artificial intelligence is here!" The creator of ELIZA was so disgusted by this that he spent the rest of his career trying to fight against the concept of machines thinking because it was being used to sell anything and everything during his time. Consider this when looking at LLMs and AI models today which are being pushed as being "agents," replacements for people. It's been done before, throughout all of the history of AI, and it'll be done again!
What would eventually become AI started by humans designing complicated but explicit instructions for how to interpret input and generate output. At their most basic, these instructions, or rules, take the form "if X, then Y." It's an elaborate flow chart. "If a traffic light is green, then cars can go." "If text includes the phrase 'Dear So-and-so', then it is a letter or email." "If someone listens to Lana Del Ray, then they may also like Billie Eilish." These sorts of rule sets can create very useful tools, but eventually the possibility space gets so complex that it becomes too much for people to be able to program each individual possibility. This is what AI was in theory created to solve.
AI technology is used daily in things you may not realize. Spell and grammar check uses AI style algorithms. Autofocus in cameras uses similar tech as image generators to figure out where the faces are to get the perfect shot. Televisions and game systems use upscalers to give you more impressive visuals with less energy use and processing power. Your health insurance company has likely used an algorithm to deny your insurance claim! This stuff is not new, but the power and extent it can be applied is.
So AI: it's new, but it's not. That's why it's so hard to track through history!
To look at a more specific Timeline of AI, let's look at some resources that take different angles on the question.
AI-Timeline.org - A robust resource distributed under the MIT Open Source license, the AI Timeline gives detailed, sourced news going back to 2015 in the area of AI
History of Artificial Intelligence - Illinois Eastern Community Colleges focuses purely on mapping the history of LLM model development.
Timeline of Major AI Events - University of South Florida gives us a succinct timeline that focuses purely on the computational power of AI models.
The History of AI - IBM focuses on the research history of AI in a detailed write-up.