Playing with a Meeting AI

“I’m going to bring an elephant to our meeting next week.”

What was that?!?

*****************************************************************************

Every week I have a call on AI: what’s in the news, how people are using it, how regular people can use it at work, play with AI tools, and any rants.

After numerous articles about how hiring managers were looking for people with AI skills, even those outside of the tech arena such as finance and marketing, I was wondering what kind of AI skills are these managers looking for? Machine learning, algorithms or something along the lines of prompting (which in my eyes is really just setting up a context and asking a question in a box). The news articles were very vague and I wanted, needed, specifics.

Would my playing around in CoPilot for Edge or in chatGPT suffice? Or do I need to be able to train a model?

If a software product had some elements of AI, would I have to prove my capability to pick up on it or do I have to have experience with said product, even if it came out just a year ago?

Do I need to get the ultimate answer after inputting a single prompt or can I play around with it and guide it to some answers or curiosities I’m looking for?

It felt like news articles were scaring us needlessly with portentous sayings such as “don’t let it pass you by” or ” those who will take your jobs will be those who know AI”.

When I first started working, there was a new product out called Lotus 123 and PCs were a new thing. In college we didn’t have computers or PCs. My first job out of college did not require familiarity with Lotus or even computers.

But on my first day at my job, there was a new computer sitting at my desk and loaded into the computer was Lotus. My lack of skills did not impede me; they hired me because I had a college degree, showed I could think and had some accounting courses.

I sat down at the desk and figured out how to use those tools. Thank you very much.

There was no consternation about being left behind because I was not taught how to use a computer or Lotus.

Same thing with the internet: no drama about how we were going to be left behind because of our lack of skills in using the internet. We just got on with the business and explored the new technology.

But today, everything has to have the “end of the world” tone unless you get trained up on it.

Anyway, that’s my rant and the fearmongering from the reporters instigated me to start a weekly call on how we could use AI – the free version of course. I basically wanted to “steal” ideas from others who may be using AI at work just to stay on top of things and being able to say, yes, I know how to use AI or learn how to use it.

During one of these calls, we began to play with the Zoom AI which led to the “bring elephant” comment. We started out doing typical explorations such as playing with the prompts provided to us by Zoom: “has our name been mentioned in the call”, “catch me up”, “what are the action items”.

We played with the names by adding our names to the conversation at such ridiculous levels that it was funny. At one point, the more creative person said, “you know, it makes me feel very accepted and recognized to have you mention my name a couple of times.”

During one of our “catch me up” episodes, the Zoom AI picked up on the good feelings due to using our names during the course of our conversation.

Then, “the AI disgusts me.” That surprising outburst made me laugh. What a change in the conversation and the AI later picked up on that comment. Later on, when we asked it to catch us up, it noted that comment but also noted that we were being very playful and exploratory and so that comment was not a serious comment. It might have been a joke.

Hmmm, interesting that the AI appeared to recognize that “disgust” statement was more of an exploratory nature rather than a serious comment. I found that interesting. Now I was laughing, so we decided to try it again but be more serious. In other words, I had to tamp down my laughter.

And the AI noted that the atmosphere got serious when the topic of AI disgusting us came up again. It was no longer an exploratory or joking matter.

So, the Zoom AI can’t see but it uses as cues noises such as laughter and maybe even tone of voice to decide whether the conversation is really on the humorous side or actually serious. It would be interesting to explore Teams or chats on whether it picks up the mood in the meetings (“what was the mood in the meeting?”) or ask who spoke a lot or who did not participate at all. This could provide interesting cues on the mood of employees or whether they are engaged.

Towards the end, the line about bringing an elephant to our next meeting was made. By that time, I was prepared for these surprises, so I did not laugh…I just went along with the conversation.

How did the AI handle this during the next “catch me up” question? It said the elephant piece was a joke.

Its reasoning was that elephants don’t typically go to meetings.

Hmmm, interesting, very interesting.

*****************************************************************************

What the AI provided during the call appeared to work, even the action item list was correct.

But I’m sorry to say the summarization of the call sent via email was not good. The summarization was like a fun house of mirrors reflecting back a distorted view of what happened. That part needs some work.