2022 was the year AI stopped being the thing research teams talked about at conferences and became the thing everyone used. GitHub Copilot went from preview to production, image generation went from technical demo to "anyone can make a picture of a cat astronaut", and then ChatGPT landed and changed everything overnight. By November, your mom had heard of ChatGPT.
GitHub Copilot's impact on developers
GitHub Copilot moved from technical preview to GA in June 2022 at $10/month. The adoption curve was steep: developers who tried it rarely turned it off. The productivity gain on routine code was immediately apparent. The more important effect was what it demonstrated about AI-assisted development: the barrier between idea and working code dropped for well-understood patterns.
The image generation explosion
DALL-E 2 launched publicly in the summer. Midjourney's Discord community grew to millions of users. Stable Diffusion went fully open source in August, putting state-of-the-art image generation on consumer hardware. By year end, AI-generated images were mainstream. The creative and copyright implications are still being worked out but the capability is irreversible.
ChatGPT as the consumer moment
ChatGPT launched November 30th and reached 1 million users in five days. It demonstrated that a language model with sufficient quality and a simple interface could be a mainstream consumer product. The product teams watching those growth numbers scrambled their 2023 roadmaps in December 2022.
What 2023 will be about
2022 demonstrated capability. 2023 will be about integration: LLMs in enterprise software, AI coding tools in developer workflows, and the regulatory and copyright frameworks catching up to the capability curve.