Companies and organisations need a clear etiquette for online meetings, when it comes to the attendance of AI bots instead of real persons.
Recently, I attended an online meeting where two persons did not turn up but sent their AI note-taker bots instead. The leader of the meeting asked everybody whether it was okay, and we decided it was not okay. There are many good reasons to establish a clear etiquette around this, so meeting attendants don’t have to waste time and decide this on the go.
In the wake of the hyped Generative AI wave, many new note-taker tools have been launched. Among the Europeans are Trint from the UK, Amberscript out of the Netherlands, and HappyScribe from France. They all integrate with Zoom and Amberscript also with Teams. Among the popular US tools are otter.ai, read.ai and fireflies.ai. (I have not tested any of them on GDPR compliance or ethics). It is convenient and easy to have tools transcribing meetings but there are many risks and ifs, as far as I see it.
- They don’t record the full meeting, they make resumes, and there is NO guarantee that they are correct – they might hallucinate, right
- If you accept more than one AI note taking tool in the meeting, there might be different versions of the meeting
- How many blank squares in a meeting do you want compared to real humans?
- The tools only participate passively – they don’t contribute anything to the meeting!
- You probably have no idea what personal data the different bots take out of the meeting, where the data is stored and what is the risk of abuse.
So, if you want to give your staff and colleagues the possibility of using AI note-taking bots, it could be an idea considering following etiquette/rules:
- Alway inform the participants who is doing AI notes and get consent. Also inform them that they can always demand that the bot is turned off
- Make sure that the tool used is GDPR-compliant and data is stored safely
- Accept only one note-taking tool in each meeting, and have the meeting leader edit it and accept it, then circulate it among the participants, so they have the same perception of the meeting. It is more important to agree on the meeting’s decisions than each participant can get their own personalised version of the meeting
- With only one AI bot in every meeting, you avoid the risk of AI bots taking over the screen and make sure that there is real dialogue at the meeting among humans
- Make sure to have a few but important meetings, so that humans want to attend and avoid that the same persons send their AI bot again and again
- Turn off the bot when you talk about confidential stuff (it also makes more people want to come) and whenever a participant asks for it
- Don’t publish any transcriptions or summaries of meetings without consent from all the participants.
According to this report from Norway, which has tested CoPilot in organisations, some people feel uncomfortable with recordings and AI bots at meetings. It might cool off after some use, especially if you are open about what you do and don’t do.
Photo: Unsplash.com
If you have more any more ideas or input to this topic, we’d love to hear from you at info@dataethics.eu
