LLMs have considerably reduced the time taken to develop a text classifier from 3 weeks to just a minute, providing a massive 7,200x productivity boost.
Despite not collecting any data in the 1-minute process, the performance of the classifier can be evaluated efficiently using an active learning approach.