Ask Your People Before You Automate

Most automation decisions are made by looking at a process from the outside and picking the steps that appear repetitive. That is a reasonable place to start. But it misses something important, which is what the people doing the work would actually like to hand over.

In a Stanford study, researchers asked 1,500 workers across 104 occupations which of their tasks they wanted AI to take on, then compared those answers against what the technology can currently do.¹ The results sorted into useful categories. Some tasks are wanted and technically ready, which makes them obvious places to begin. Others are technically ready but unwanted, and automating those reliably produces the resistance that stalls adoption. Workers also showed a consistent preference for AI as a partner rather than a replacement, and the strongest motivation they gave was freeing time for more valuable work.

You can run a small version of this in your own business in a week. Ask each person on a team to list the tasks they actually did over the past month, then rate each one on three things: roughly how much time it takes, how much they enjoy doing it on a scale of one to five, and to what degree they believe the task can be automated.

Putting those three together brings some clarity. Tasks that take considerable time, score low on enjoyment and look automatable are your starting list. Tasks people enjoy might be worth leaving alone even when an AI tool could do them, because that is frequently the part of the job that keeps a good employee.

This exercise changes the conversation about adopting AI. An abstract fear becomes a specific and usually reassuring discussion, and you end up with a better list than someone from outside the workflow has to rely on their external observations.

It also works as a measuring stick. Run it again once the automation is live. If the enjoyment scores have fallen, you most likely have automated the wrong part of the job.

Sources

¹ Future of Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce, Stanford University, 2026, based on responses from 1,500 workers across 104 occupations and annotations from 52 AI experts.