Today, I want to talk about AI and meeting notes.
A lot of us now use AI to take meeting notes. Why do so many people want a record of what happened after a meeting? Because, finally, there is something that can act as a secretary for us: take notes, capture the content, and turn what was said into a summary.
But what I really want to say today is that people use meeting notes in very different ways.
Take a hypothetical example. We all talk about how AI can transcribe a meeting, summarize it, and pull out action items. But there is a problem. Someone might say during the meeting, “Let’s take another look at this next week.” The notes might turn that into, “We’ll finish this next week.”
There is an important distinction here: discussing something does not mean you have committed to doing it. This is one of the problems AI can run into when producing meeting notes.
So I think we can put something like this into the prompt. When we were working on prompt templates before, we thought about this too. First, we need to establish which things in the meeting have been decided and which have not. Who is responsible? Has a time been clearly stated?
If something has not been made clear, mark it as “to be confirmed.” Don’t let the AI fill in the gaps itself just to make the meeting notes look complete.
You see, different people use AI for meeting notes differently. After hearing what I’ve just said, you already have a different idea of how to use it. The way you use AI to process those notes changes the efficiency you get out of it.
For example, does getting a voice recorder for meetings mean I can now run better meetings? No, it doesn’t.
First, we’ve recorded a lot of content from a meeting. But why do we have one meeting and then another? There must be issues that haven’t been properly resolved, or parts of the process that aren’t working efficiently. I think we can look at it this way.
There is communication between people, communication between people and AI, and communication between AI systems. My view is that AI-to-AI communication is the most efficient, and that communication between a person and AI is more efficient than communication between people.
But there is a qualification here. In some situations, if the AI doesn’t understand your background, people communicating directly may understand each other better and communicate more efficiently than they would with AI involved.
In the meeting scenario I’m talking about, I think what really matters is how we use AI to improve the efficiency of decision-making and fill in missing information.
Some companies prepare their meetings very thoroughly. Before a meeting, they get the material ready: here is item one, item two, item three. The things they want to discuss are prepared in advance.
In other cases, people don’t prepare like that. Something happens, and we say, “Come over, let’s talk about it.” Two people talk it through, make a judgment, and go and do it. That can be relatively simple.
But why do so many meetings need materials prepared beforehand?
The same question applies when you give your meeting notes to AI. The background information it has is bound to be incomplete.
So when we work on meeting notes, and when we look at them ourselves, I think one very important thing is the information behind those notes.
