How much time does AI save at work? New Census Bureau data breaks it down.

Artificial intelligence is believed to save workers time by helping them complete their daily tasks. But just how much more efficient can it make employees?

About 55% of workers say they use AI on the job, with about one-third of those reporting it cuts the time they need to complete a given assignment by one to two hours, according to new data from the U.S. Census Bureau.

Workers reported turning to AI for help with tasks including writing, generating new ideas, conducting research and completing administrative tasks.

A smaller share of workers use it to write code, for medical care and to manage logistics or supply chains, according to the Census data.

The findings are based on a weekly national Census Bureau survey of individuals and households on topics including employment, food and nutrition, and transportation.

One-quarter of employees who reported using AI at work say the tools shaved less than an hour off the time it would otherwise have taken them to complete a task, according to the Census data.

A smaller share, at 15%, of workers saw even greater efficiency gains, saying that AI cut their workloads by three hours per task. An additional 15% said it saved them four hours.

To be sure, it can take time for employees to learn to use new technology like AI, which can initially add hours to the time required to complete a task. Research from the Massachusetts Institute of Technology shows that AI adoption in manufacturing initially decreases productivity before ultimately helping workers achieve long-term gains.

Economists use the "J-curve" model to explain the phenomenon. The productivity trajectory, in the shape of the letter J, reflects the period when considerable resources are committed to learning and investing in new technologies, with productivity initially dipping before rising exponentially.