The tools arrived faster than any language for describing what they changed about the work itself.
The phrase "AI-assisted work" entered widespread use in the early 2020s as large language models, image generators, and code-completion tools moved from research laboratories into daily professional practice.1 The underlying technology had been developing for decades, from the neural networks of the 1980s to the deep learning breakthroughs of the 2010s, but the public inflection point came in November 2022 with the release of ChatGPT by OpenAI.
Within two months of its launch, ChatGPT reached an estimated one hundred million users, making it one of the fastest-adopted consumer applications in history.2 GitHub Copilot, a code-completion tool developed by GitHub and OpenAI, reported that developers accepted roughly thirty percent of its suggested code completions by 2023.3
A 2023 study by researchers at OpenAI and the University of Pennsylvania estimated that roughly eighty percent of the U.S. workforce could have at least ten percent of their tasks affected by large language models, and approximately nineteen percent could see at least fifty percent of their tasks affected.4 The occupations most exposed included interpreters, writers, tax preparers, and mathematicians, roles that had long been considered resistant to automation.
The International Labour Organization estimated in 2023 that generative AI was more likely to augment existing jobs than eliminate them, with clerical work identified as the most exposed category globally.5