AI meeting notes aren't making employees forget how to listen, at least not on the evidence published so far. They can make it easier to stop paying attention, though, and that is the real risk. There is research suggesting people put in less critical effort when they trust an AI tool. There is also older research showing that the way we take notes affects what we remember. The sensible position sits between "AI notes ruin attention" and "AI notes free us to think". It depends on how teams use them.
This opinion piece weighs both sides, looks at what the research does and doesn't show, and suggests meeting habits that keep the benefits without the passive-listening trap. For a related debate on AI and workplace habits, see AI Agents Are Getting Smarter. Are Your Employees Getting Lazier?
What is the worry, exactly?
The worry is simple. If a tool promises to capture "who said what" and list the action items, some attendees will stop tracking the discussion themselves. They'll answer emails with the meeting running in the background and read the recap later. The meeting goes on, but fewer people in it are actually thinking.
The feature set makes this tempting. Microsoft describes Copilot in Teams as able to summarise key discussion points, show where people agree or disagree, and suggest follow-up tasks in real time. Microsoft's support pages also say Copilot answers questions about the meeting in real time. Zoom's AI Companion can email a summary after the meeting if the host chooses to share it. These are useful features, and they also make it easier to stop listening.
What does the research actually say?
There is no large study we know of that measures whether AI meeting notes specifically reduce listening skill over time. Anyone claiming otherwise should show their source. What exists is related research worth knowing about.
| Study | What it looked at | What it found | What it means for meetings |
|---|---|---|---|
| Lee et al., Microsoft Research, CHI 2025 | Survey of 319 knowledge workers sharing 936 examples of using generative AI at work | Higher confidence in the AI was linked to less critical thinking; higher confidence in one's own ability was linked to more | Trusting the recap too much may reduce the effort people put into checking and thinking |
| Mueller and Oppenheimer, Psychological Science, 2014 | Students taking lecture notes on laptops versus by hand | Laptop note-takers tended to transcribe verbatim and did worse on conceptual questions | Processing ideas in your own words may matter more than capturing everything |
| Morehead, Dunlosky and Rawson, Educational Psychology Review, 2019 | Direct replication of the 2014 study | Did not reproduce the conceptual-question advantage; note-taking method mattered little | Be careful drawing strong conclusions from a single study |
Taken together, the research is suggestive, not conclusive. The Microsoft study is self-reported survey data, not a measure of listening ability. The note-taking studies involved students and lectures, not colleagues in a working meeting. Still, the common thread is plausible. Engagement depends on whether people actively process what they hear, and tools that do the processing for them can lower that effort.
What is the case that AI notes help people listen?
There is a fair argument on the other side. Taking manual notes is itself a distraction. Someone typing minutes is half-listening while they write. Handing the record to software can free attendees to look up, ask questions and respond.
AI notes also help people the old approach let down:
- Accessibility. Live transcripts help colleagues who are deaf or hard of hearing, or who are following a meeting in their second language. That covers many people in Singapore's multilingual workplaces.
- Remote attendees. People on poor connections or in different time zones can catch up on what they missed.
- Fairer credit. A transcript with speaker names makes it harder for a quieter colleague's idea to be absorbed into someone else's summary.
- Less admin. Nobody has to be the permanent minute-taker, a job that often falls to the same junior staff.
So the tool itself isn't the problem. The problem is using it as a substitute for being present rather than as a support for it.
Where does it go wrong in practice?
From the published research and vendor documentation, a few patterns are worth watching for:
- Multitasking by default. If everyone assumes the recap will cover it, attention drifts. The meeting becomes a broadcast rather than a discussion.
- Reading instead of remembering. People who never engaged with a discussion can read a summary of the outcome. That doesn't tell them the reasoning, the objections or the tone.
- Unchecked summaries. Microsoft itself warns that some AI-generated recap content might be inaccurate or incomplete. People who weren't listening are the least able to spot mistakes. Our companion post Your AI Meeting Summary Might Be Wrong. Who Takes Responsibility? looks at the accountability side.
- Too many meetings. If AI recaps let people skip meetings that should never have been held, that is a good thing. If they let organisers keep adding attendees who will never speak, the meeting just gets bigger and less useful.
How can teams keep the benefits without losing attention?
The fix is mostly about habits, not software settings:
- State the purpose first. Open with the decision the meeting needs to reach. People listen better when they know what they are listening for.
- Use AI for the record, not the thinking. Let the tool capture details. Ask attendees to note in their own words the one or two points that affect their work.
- Close with a human recap. Someone summarises decisions and owners out loud before the meeting ends. It reinforces memory and gives the AI a clean summary to work from.
- Cameras and devices on purpose. In small groups, agree that laptops are for the shared screen, not inboxes.
- Invite fewer people, share the recap more widely. Keep active participants small, and send the reviewed summary to those who only need to know the outcome.
- Review before relying. The organiser checks the AI summary before it goes out.
Good audio helps here too. When people can hear clearly, they follow more easily, and the transcript is better. For individuals in open-plan offices or at home, a headset like the Jabra Evolve2 85 MS Stereo ANC Wireless Headset (S$635.00) or the HP Poly Voyager Focus 2 Microsoft Teams Headset (S$314.05) cuts out surrounding noise. For small rooms, a speakerphone such as the Jabra Speak2 55 MS Wireless Conference Speakerphone (S$185.00) puts a proper microphone in the middle of the table. Prices as listed on 25 Oct 2026.
So, are AI notes making us worse listeners?
Our view is this: AI notes don't erode listening on their own, but they remove a reason to listen. Organisations that treat the recap as permission to disengage will probably see shallower meetings. Those that treat it as a safety net, while still expecting people to be present, get the time savings and accessibility gains without that downside.
If you are rolling these tools out, decide early what good meeting behaviour looks like and say it plainly. Before switching AI recaps on across the company, also make sure people know how and when meetings are recorded. See Should Your Meeting Room Be Allowed to Record Everything You Say? and the policy guide Does Your Company Need a Policy Against AI Recording Devices?
What to do next
- Agree a short meeting etiquette note: purpose first, verbal recap last, laptops for shared content only.
- Decide which meetings get AI summaries and who reviews them.
- Ask staff for feedback after a month. Are meetings shorter, or just quieter?
- Fix audio in the rooms and on the desks where it is poorest.
- Share reviewed summaries widely, so fewer people need to attend just to stay informed.
Frequently asked questions
Is there evidence that AI meeting notes reduce listening skills? We are not aware of a study that measures this directly. Related research, including a 2025 Microsoft Research survey, links higher confidence in AI tools with less critical thinking effort, which suggests over-reliance is a risk worth managing.
Is handwritten note-taking better than typing? The evidence is mixed. A 2014 study found laptop note-takers did worse on conceptual questions, but a 2019 direct replication did not reproduce that result. Processing ideas in your own words appears to matter more than the tool.
Can AI meeting notes improve accessibility? Yes. Live transcripts and summaries help colleagues who are deaf or hard of hearing, people working in a second language and remote attendees who miss parts of a call.
Should we ban AI notes to make people pay attention? A ban is unlikely to fix attention and removes real benefits. Clear meeting habits, smaller attendee lists and a human recap at the end tend to work better.
Does better audio help AI notes and listening? It helps both. Clearer audio makes speech easier for people to follow and gives speech-to-text a better input, which improves the transcript that summaries are built from.
