[herculity275]

* 2017 - the Attention Is All You Need paper proposes transformers, suggests language translation as their primary application

* 2019 - GPT2 is presented to the public and demonstrates transformer-based LLMs and their emergent capabilities

* 2020 - GPT3 is released and shows that throwing more compute at LLMs yields significantly better LLMs

* 2022 - ChatGPT is openly released to the public, showcasing the versatility of an LLM-based chatbot

In my experience transformers have been all the rage in the researcher/enthusiast scene since 2019. The technology has just gradually matured enough to become viable for consumer use, which is why you see the industry rushing to adopt it. ChatGPT was the watershed moment for the tech because suddenly anyone in the world could sign up for free, open a chat dialogue and start getting legible LLM output without needing to understand the tech or prompt engineering.

[jerpint]

Not to mention without needing expertise to deploy the thing