4 articles.
The most-cited evidence on AI and work productivity comes from 5,179 customer support agents and found a 14% average gain in issues resolved per hour, but the gain was almost entirely concentrated in novice and low-skilled workers while experienced staff saw almost nothing. The effect was not a uniform speedup, which matters far more for hiring, training and retention decisions than the headline average.
METR ran a randomised controlled trial where 16 experienced open-source developers worked on issues in repositories they knew, with AI tools freely available in half the tasks. Implementation took 19% longer with AI. The developers predicted a 24% speedup before starting and still estimated a 20% speedup afterwards. Economics experts had forecast 39% and ML experts 38%. No confidence interval was published for the headline number, and the study has since been superseded.
Noy and Zhang's large field experiment found that generative AI cut time on mid-level professional writing tasks by about 40% and raised quality ratings, but the benefit depended on task type and on workers' initial skill.
Before you split a group subscription for your next presentation: Beautiful.ai's own pricing pages disagree on the Pro price and on how many seats Team includes, Gamma publishes exactly what its PowerPoint export loses, and the presentation tool that reached 20 million users had its founders walk away to build something else entirely. Pricing a five-person group properly, and testing the handoff, is the whole lesson.