Say where AI helped—and who can explain the result

What we found in the news

A July 2025 Journal of Participatory Medicine study (opens in a new tab) explored patients’ views of AI in several health-care scenarios. In discussing transparency, the authors described requests for:

“clear explanations of what contributions AI made”

— Garrett Foresman and colleagues

The study involved a small co-design session and considered clinical examples. It does not show that a disclosure label improves outcomes. It does invite a more useful explanation than simply announcing that AI was involved.

Why this matters to us

The phrase AI-assisted can cover very different work. It might mean help choosing a heading, preparing a first draft, summarizing information, or suggesting an action. A reader cannot tell which from the label alone.

Imagine a staff colleague showing a participant a draft and final activity invitation. The participant asks about a changed detail concerning the meeting place.

The colleague can explain which part used assistance, where the confirmed location came from, and who checked the final version. The conversation stays focused on the information the participant needs.

At genus, we want explanations to help people understand responsibility and ask a useful question about the result in front of them.

Describe the contribution at the right level

Choose language that fits the task. If AI helped draft wording from approved details, say that. If it proposed a summary for someone to check, describe the review too.

Avoid making the explanation larger than the actual use. A minor editing aid does not mean the whole service is automated, while substantial assistance should not disappear behind vague wording.

In our imagined invitation, the source of the meeting place matters. The team should be able to distinguish a confirmed program detail from wording the tool suggested.

Offer more information where it helps someone make a decision. A short explanation can be followed by a clear route to ask about the particular process or source.

Name the responsibility around the result

A tool’s contribution does not tell the reader who can answer a question. Make the relevant person or service contact understandable.

The colleague reviewing the invitation should know which details they are responsible for checking. If another team supplies the location or access information, there should be a route to confirm it.

Give the recipient a practical way to raise a concern. A participant who notices the wrong entrance needs help correcting that detail; a generic page describing AI principles may not solve the immediate problem.

Where uncertainty remains, say what will happen next. Someone should be able to distinguish a confirmed answer from a question awaiting review.

Try the explanation with the person receiving it

Ask a few people to read the explanation and describe what they think happened. Their responses can reveal ambiguity the writer did not anticipate.

One person may think the AI has read their private information. Another may assume that a human checked every sentence. Those interpretations deserve attention if they do not match the actual workflow.

Use fictional material or suitable public content when testing the explanation. The team can learn about clarity without exposing someone’s private communication.

Revise the wording where needed and keep it aligned with changes to the service. If the tool’s role expands, an old explanation may no longer be accurate.

The point is not to persuade everyone to feel the same way. It is to give people a truthful basis for understanding the work and deciding what they want to ask.

What this means for the people you serve

People can understand where assistance entered the process and who remains responsible for the communication they receive. They should have a way to question a detail without needing to understand the technology behind it. Clear disclosure can support that conversation when it connects to real information and reachable people.

The genus connection

Today, genus partner-portal workflows can include bounded AI assistance, staff review, and preparation for an intended audience. The precise contribution depends on the enabled function and the workflow being used.

We would describe that contribution alongside the review responsibility. Confirm the source of important details, prepare an understandable explanation, and give people a suitable contact for questions. The explanation should reflect what the team actually does with the tool.

One question to take away

Could someone reading this result tell where AI helped and who can explain the detail they are questioning?