Summary of MIT research (march 2023)

The paper from Shakked Noy and Whitney Zhang from MIT examines the productivity effects of a generative artificial intelligence technology called ChatGPT in the context of mid-level professional writing tasks. In a pre-registered online experiment, 444 college-educated professionals were assigned occupation-specific, incentivized writing tasks, and half of them were randomly exposed to ChatGPT.

The study found that ChatGPT substantially raises average productivity and output quality in mid-level professional writing tasks. The technology compresses the productivity distribution by benefiting low-ability workers more and mostly substitutes for worker effort rather than complementing worker skills. ChatGPT restructures tasks towards idea-generation and editing and away from rough-drafting. Exposure to ChatGPT increases job satisfaction and self-efficacy and heightens both concern and excitement about automation technologies.

The key takeaway from this study is that generative artificial intelligence technologies like ChatGPT can have significant implications for production and labor markets. They may either displace humans completely from certain occupations or complement existing human workers and increase their productivity. In the case of ChatGPT, it has been shown to improve productivity by reducing the time taken to complete writing tasks while also improving the quality of the output. It also benefits low-ability workers more by compressing the productivity distribution. However, it mostly substitutes for worker effort rather than complementing worker skill