AI adoption survey questions to ask employees

Good AI adoption survey questions ask about real tasks: what people last used the tool for, where it helped, where it let them down and what stopped them using it again. Group the questions by purpose (current use, usefulness, training, trust and policy, worries, and what would help) and ask each one openly, with a follow-up on the answer. A score from 1 to 5 tells you adoption is low. The follow-up tells you why.

Below are 24 questions you can copy. Pick five to eight for a short round. Asking all of them at once makes for a long survey and tired answers.

What should I ask employees about our AI rollout?

Start with what people actually do, then move to how they feel about it. If you lead with “How do you feel about AI?” you get opinions about AI in general. If you lead with “What did you last use it for?” you get an example, and the opinion follows from it.

Current use

1. Which AI tools have you used for work in the last two weeks, approved or not?
2. Tell me about the last time you used one. What was the task?
3. How often do you use it now, compared with the first month?
4. Is there a task where you tried it once and then stopped? What happened?

Usefulness for your job

1. Which part of your week would you most like help with?
2. Where has the tool saved you time? Give me an example.
3. Where has it made your work slower, or added a step?
4. How much do you have to check or correct what it gives you?

Training and support

1. What training or guidance have you had, and how much of it matched your actual work?
2. Who do you ask when the tool gives you something you can’t use?
3. Do you have time in your week to learn it? If not, what takes that time?
4. How does your manager talk about the tools?

Trust and policy clarity

1. How clear is it to you what you’re allowed to use AI for?
2. What kind of information would you never put into the tool, and why?
3. Have you held back from using it because you weren’t sure it was allowed?
4. How far do you trust its answers for your kind of work?

Worries about jobs and evaluation

1. How comfortable would you be telling your manager you used AI on an important piece of work?
2. Where you work, does using AI look like working smarter or like cutting corners?
3. What worries you, if anything, about how AI will change your role?
4. What do you think leadership wants to get out of these tools?

What would help

1. If you could change one thing about how the tools were rolled out, what would it be?
2. What would make you use it for one more task next month?
3. What have you seen a colleague do with it that you’d like to try?
4. Is there anything we haven’t asked that we should know?

Why ask about policy, training and worries at all?

Because that is where the big surveys find the gaps. In Gallup’s 2025 research, 44% of employees said their organisation had begun integrating AI, but only 22% said it had communicated a clear plan for doing so, and 30% said it had general guidelines or formal policies for using AI at work. BCG’s AI at Work 2025 survey of more than 10,600 people found that only a third of employees say they have been properly trained.

The worries are real too. In the 2024 Microsoft and LinkedIn Work Trend Index, 52% of people who use AI at work said they were reluctant to admit using it for their most important tasks, and 53% worried that using it on important work makes them look replaceable. If you skip the questions on evaluation and job security, you miss the reasons people keep their use quiet.

Why does an open follow-up beat a 1 to 5 scale here?

Take a common scale item: “The AI tool is useful for my job.” A finance team scores it 2 out of 5. That 2 could mean any of these:

• The tool can’t open the files they work in, so it doesn’t fit month-end at all.
• Someone asked it to reconcile two ledgers in week one, got a confident wrong answer, and told the rest of the team.
• They aren’t sure they are allowed to paste supplier data into it.
• They had a one-hour webinar and no time since to try anything.

Each needs a different fix: a different use case, a better first example, a clear rule, or protected time. The score can’t tell them apart. A follow-up question (“What happened the last time you tried?”) usually can, in one answer.

Scales also invite the answer people think they should give. If using AI quietly feels risky, “How often do you use AI?” gets a cautious number. Asking for the last task you used it for gets a story, and a story is harder to shade.

Keep a scale if you need a number to track over time. Put an open question next to it, and read the answers before you read the average.

How should you run an AI adoption survey?

Timing matters. Ask a few weeks after launch, once people have had a chance to try the tools on real work, then again a quarter later. The first round shows where adoption stalls. The second shows whether what you changed helped.

Make it safe to answer honestly. Tell people who will see the results and in what form. Report only for groups large enough that no one can be picked out; our post on the employee survey anonymity threshold explains how to set that number. Questions about job worries and quiet tool use get thin answers if people suspect their manager will read them with names attached.

Compare departments. A contact centre, a legal team and a warehouse office will give very different answers, and the average hides all three. Look at the usage data next to the answers: our guide to measuring Copilot adoption beyond licence data covers what the Microsoft reports show and what they leave out.

Spradley runs these questions as one-on-one AI interviews, by text or voice. Invitations go out by email or Slack, people answer on their own phone or computer, and the interviewer asks a follow-up on each answer. The report groups the answers into themes with the quotes behind them, by department, and never shows numbers or ratings for fewer than five people. See our AI adoption page for how a round works.

What do you do with the answers?

Sort them by the reason behind them. You will usually find a short list of reasons that explain most of the low use, and they differ by team. We go through the most common ones in why employees don’t use AI tools. If the current-use questions turn up personal accounts and unapproved apps, read our post on shadow AI before you reach for a ban.

Then tell people what you heard and what you will change. A second round gets better answers when the first one led to something visible, such as a clearer policy, a team session built on their own tasks, or a tool switched off because it added a step.

Frequently asked questions

Key terms

open question

A question people answer in their own words rather than on a scale, such as “What did you last use it for?”

follow-up question

A question asked in response to a specific answer, to get the example or reason behind it.

shadow AI

Employees using AI tools for work that their organisation has not approved, such as a personal chatbot account, a browser extension or an app on their own phone.

minimum group size

The smallest number of responses a group needs before its results are shown. Five is a common default; smaller groups are hidden or combined with a larger one.

AI-moderated interview

A one-on-one conversation run by software, in which the interviewer asks open questions and a follow-up question based on each answer. Everyone invited can take part at the same time.