The idea of artificial intelligence dates back to Alan Turing in the 1950s and the Dartmouth Conference. For decades the field oscillated between "AI winters" and booms. In the 2010s deep learning — especially AlexNet in 2012 for vision and the Transformer architecture in 2017 — reignited it. But the output of these models mostly stayed in technical papers; it did not touch everyday life.

2022 changed all of that. In July, Midjourney opened its beta; in August, Stable Diffusion was released as open source. Suddenly anyone could write a paragraph of text and get photorealistic images back in minutes. The same year, an image generated with Midjourney — 'Théâtre D'opéra Spatial' — won the digital art prize at the Colorado State Fair, and the art world erupted into debate.

On 30 November 2022, OpenAI made a chat interface built on GPT-3.5 — ChatGPT — freely available to everyone. It hit a million users in five days and a hundred million in two months, the fastest growth ever recorded for a consumer technology. Users wrote essays, generated code, did homework, had therapy conversations, brainstormed ideas. Google and Microsoft scrambled to release their own models; how to assign homework in classrooms, journalism ethics, copyright law, and the labour market were suddenly all reopened.

Anthropic's Claude, Meta's Llama, Google's Gemini, xAI's Grok joined the competition through 2023–2025, and models grew markedly more capable each year. By 2025 these models were taking on a significant share of software engineering, the bulk of customer service, and almost all translation and summarisation. Whether the long-term impact will rise to the level of electricity, the internet, or the printing press is still debated; but the end of 2022 will remain a threshold in the history of AI, with a clear before and after.