Is there anything else you'd like to tell us about AI at work?

What employees should tell you about AI at work

When staff are asked, “Is there anything else you'd like to tell us about AI at work?”, the useful answers are rarely just enthusiasm or fear. The real value is in surfacing where people are already using AI, where they feel exposed, and which workflows need clearer rules before adoption becomes invisible.

The source question for this draft came from: AI governance guidance for UK SMEs. For smaller regulated or professional-services teams, the point is practical: you need enough candour to govern real AI use, not a policy that only describes approved tools on paper.

The best answer is honest evidence of use, risk, and support needed

Employees should tell leaders where AI is already helping, where it is being used unofficially, and where they are unsure about data, accuracy, or client impact. That gives the business a realistic AI map: tools in use, workflows affected, sensitive data risks, and training gaps.

A good response might say: “We use AI to summarise meeting notes and draft internal documents, but people are unsure whether client data can be pasted into public tools. We need approved tools, examples of safe use, and a simple review process for anything client-facing.”

Find the AI use your policy is currently missing

What leaders should listen for

Do not treat this question as a comments box exercise. It should reveal operational signals that a written policy may miss. Look for patterns in where employees are experimenting, where they are blocked, and where they are making judgement calls without guidance.

• Shadow AI: teams using personal accounts or unapproved tools because approved options are slow or unclear.
• Data uncertainty: staff unsure what client, employee, or commercial data can be entered into AI systems.
• Quality concerns: outputs that look polished but need checking for facts, context, bias, or regulatory fit.
• Workflow pressure: managers expecting AI productivity gains without changing review, workload, or accountability.
• Training gaps: employees wanting practical examples rather than abstract policy language.

How to turn feedback into governance

The feedback should become a governed action list. Start by grouping responses into approved use cases, risky use cases, banned use cases, and unclear cases. Then assign owners for each unclear workflow and decide whether it needs a tool change, training, review step, or prohibition.

This is where many AI policies fail. They ask for responsible use, but they do not create a route for employees to report messy reality. If staff cannot safely say “this is already happening”, leaders will govern a fiction.

Create an AI governance loop that staff will actually use

The minimum evidence to keep

Keep a lightweight record of what was reported, what decision was made, and who owns the follow-up. For example: “marketing uses AI for first-draft blog outlines; no client data entered; human review required before publication; owner: marketing lead; next review: quarterly.”

This gives the business an audit trail without burying teams in bureaucracy. It also helps leaders see whether AI adoption is improving work, shifting risk, or simply creating hidden review debt.

A practical question set for SMEs

If the open-ended question produces thin answers, ask five follow-ups. Which AI tools are you using? What data do you enter? What work does AI influence before a client, customer, or colleague sees it? What are you unsure about? What would make safe use easier?

Those questions turn vague sentiment into governable facts. They also show employees that the goal is not to catch people out. The goal is to make useful AI adoption safe, visible, and repeatable.

Build practical AI controls into day-to-day workflows

Conclusion

The most useful thing employees can say about AI at work is what is actually happening. Leaders need to hear where AI is useful, where it is risky, and where policy has not caught up. That evidence is the starting point for credible AI governance.