AI & Enterprise
LLM withdrawal study shows knowledge workers\' reliance on AI tools
A study by a Korea Advanced Institute of Science and Technology (KAIST) team found that temporarily cutting off access to large language model tools such as ChatGPT can significantly disrupt office work. In a four-day ban involving 10 frequent users, participants reported stress, delays and a sense of loss, and some postponed tasks or lowered standards. The study also found some regained clarity and ownership by designing reasoning themselves, while warning LLM use is becoming a social norm and infrastructure.
