Hong-Yi TuYe

In a preregistered online experiment with 1,893 participants who submitted more than 18,000 prompts and generated over 300,000 images, users attempted to replicate a target image in 10 tries using one of three randomly assigned models: DALL·E 2, DALL·E 3, or DALL·E 3 with automated prompt rewriting. Users with access to DALL·E 3 achieved higher image similarity than those with DALL·E 2—but only about half of this gain (51%) came from the model itself; the other half (49%) resulted from users adapting their prompts.

Prompt adaptation emerged across the skill distribution, was driven by trial-and-error, and could not be replicated by automated prompt rewriting, which erased 58% of the performance improvement associated with DALL·E 3. The findings position prompt adaptation as a dynamic complement to generative AI and suggest that without it, a substantial share of the economic value created when models advance may go unrealized.

Remote work surged during the Covid pandemic, but published estimates disagree about the magnitude of the shift. Using the nationally representative Remote Life Survey fielded in October 2020, we document that 31.6% of continuously employed workers always worked from home, 21.9% worked from home sometimes or rarely, and 53.5% worked from home at least one day per week.

We benchmark these results against government surveys and quantify how four measurement choices drive discrepancies: (a) mode of survey administration, (b) whether self-employed workers are included, (c) differences in industry mix, and (d) whether surveys exclude respondents who were already remote before the pandemic. The last factor explains most of the gap between the Current Population Survey and other measures. Under our preferred adjustments, about half of the U.S. workforce was working remotely at least one day per week in December 2020.

We report the results of a nationally representative survey conducted in two waves (April 1–5 and May 2–8, 2020) to understand how COVID-19 reshaped US employment. Among individuals employed before the pandemic, about half were working from home by early May, including 35.2% who had recently shifted from commuting. Another 10.1% reported being laid off or furloughed.

States with higher COVID-19 incidence saw more workers switch to remote work, while states with greater shares of information-intensive occupations experienced larger shifts to working from home and fewer layoffs. Younger workers were more likely to transition to remote arrangements. We observe little change between the survey waves, indicating that most remote-work adjustments occurred by early April 2020.