An early look at the labor market impact potential of large language models

We examine the potential implications of Generative Pre-trained Transformer (GPT) fashions and associated applied sciences on the U.S. labor market. Utilizing a brand new rubric, we assess occupations primarily based on their correspondence with GPT capabilities, incorporating each human experience and classifications from GPT-4. Our findings point out that roughly 80% of the U.S. workforce might have at the very least 10% of their work duties affected by the introduction of GPTs, whereas round 19% of staff may even see at the very least 50% of their duties impacted. The affect spans all wage ranges, with higher-income jobs probably dealing with higher publicity. Notably, the impression just isn’t restricted to industries with greater current productiveness progress. We conclude that Generative Pre-trained Transformers exhibit traits of general-purpose applied sciences (GPTs), suggesting that as these fashions might have notable financial, social, and coverage implications.