Re: RE using AI minutes

Hi all,

I agree with Gregg that replacing or augmenting human scribes with AI is a lot harder than it seems.

As someone who scribes regularly, I don't see human scribing as a problem, I would also happily offer to summarise stuff. Even if human mistakes sometimes happen, I don't think this is a problem well solved by automating to something that also makes mistakes. 

Sure I'm excited about new tech as the next person, but I want to express my concerns (sorry to point out some elephants in the room):

1. Ethics - major large language models rely on stolen training data, and they use low wage workers to 'train' at the expense of the well being of those workers. 

2. Environment - Apart from raw material usage that comes with increase in processing power, LLMs uses a lot more energy and water than human scribes and summarisers do (both during training and at point of use). Magnitudes more, not negligible, such that major tech cos are building/buying nuclear power plants and areas near data centres suffer from water shortages and price hikes. Can we improve disability rights while disregarding environmental effects?

3. Quality - we've got a  lot of experts in our group: who are sometimes wrong, sure, but it seems like a disservice to their input, knowledge and expertise to pipe their speech through LLMs. From the couple of groups I've been in that used AI summaries, I've seen them:
a. miss the point a lot of the time; it looks reasonable but doesn't match up with what people said/meant; 
b. 'normalise' what was said to what most people would say, so it biases towards what's more common in training data, rather than towards the smart things individuals in this group often bring up. Normalising seems orthogonal to innovation?
c. create summaries that are either very long and wooly, with many unnecessary words, or short but incorrect. 

If we're considering if it's technically possible, I'd urge us to consider the problems with these systems too, including in ethics, environmental impact and quality.

And maybe creative ways to get more scribes, I love the trophies! 

Best,
Hidde

> On 11 Mar 2025, at 07:27, Gregg Vanderheiden RTF <gregg@raisingthefloor.org> wrote:
> 
> The topic of AI came up in F2F
> 
> to keep from using up too much time there here are some thoughts on ideas
> 
> 
> 1)  A good way to handle foreign language speaker not being as well recognized
>  - use AI to transcribe
>  - have “backup” scribe that only steps in when recognizer faile 
> 
> 2) for Off minute comments you can turn off the mike to the AI transcriber off  (see note below - where we need a special AI tool not just the one in zoom - for a number of reasons.
> 
> 3)  If we use AI summaries -we NEED to implement it in a fashion that provides a method for “in meeting” instant correction of mistakes.  In fact we need that for the human done minutes which also often misunderstand and sometimes scribe exactly the opposite of what the speaker said
> 
> 
> 4) Remember the distinction between transcript and minutes
> AI can do both but tends to do the summaries only after a meeting.  This is too late for someone to notice that the summary is not complete - is inaccurate - misinterpreted - or is opposite of what was said
> Suggestion
> Need a special tool to change the transcript into summaries on a per commenter basis - like a scribe does
> that summarization would go into the IRC
> BETTER though would be for these summarizations (minutes) to appear in a separate document that is group correctable so that the commenter can correct the AI summary. per #3 above
> 
> 5) Speaker ID 
> This is often cited as an AI shortcoming.  But it is just as much a problem in human minutes.   
> our best solution to this for both is using queue  and having chairs announce each new speaker (as they do)
> Sometimes things get into a back and forth between two people 
> Best solution is for each person to say their name before the speak each time
> ALSO 
> If we have a human  “shadow scriber” that is just monitoring the AI - they could add speaker names where the AI does not.
> 
> 6) for hybrid meetings - I  suggest we purchase more microphones so there is no-one 
> 
> 
> IF WE WERE TO USE AI  — given all the above and the comments I  think we should build (or find) one that does what we need.
>  When you think about it - it is much more complicated than it first appears  to replace or even augment the human scribe.   But it would be wonderful for the scribes once you do.
> 
> Gregg