723
Efficency
(pawb.social)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 12 Jul 2025
723 points (99.3% liked)
Fuck AI
3464 readers
647 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
founded 1 year ago
MODERATORS
I have strong doubts about this happening in the not-too-distant future just from bots 'seeing' each other in the call. Too many variables (different platforms, different times joining the call, different default languages, different preferences on how notes should be taken). This is why many platforms offer an option to use a single canonical transcription bot - this sort of thing isn't quite as easy as you seem to imagine.
For most of these calls, every participant is using the same link to join.
Each of these note taker bots is a different interface for the same AI service. The AI is not actually 25 different bots. It is 25 different faces of the same bot.
It is trivial for that one bot to recognize it has been directed to use the same link to connect to the same conference 25 times.
Yes, it is going to consolidate the work for all of those individual "faces" into a single task. It is absolutely ludicrous to assume it will be configured to duplicate its efforts 25 times.
I can absolutely see a single provider of collaborative online meetings spaces offering a 'notes' bot to the chat as a whole if they don't already do so, since that's already a thing. I see absolutely no reason why they should invest time in rearchitecting the conceptual boundaries of how private a user's interaction with an AI is. I think you've missed some of the implications of collapsing the multiple to a single whilst retaining the illusion that it is a private space
Profit.
If AIs use as much power and resources as we've been led to believe, there are massive cost savings to be had by simulating multiple bots instead of using multiple bots. If they've budgeted to earn a profit from the operation of 25 independent bots, what are they earning by running only one and claiming it is 25?
There is very little chance that this degree of optimization hasn't been employed.
For shared notes taking in a controlled environment - yes. It transparently happens anyway. For dynamic environments hosted by a separate service it's a whole other can of worms. You now seem to be fairly clear you're talking about the former, largely doable and indeed mostly implemented case. Nothing interesting left to talk about then.
Quite the reverse, actually. That "dynamic" environment hosted by a separate service is not nearly as significant as you portray it. The entire point of a meeting is for every observer to share the same experience.
Again, it is completely trivial for the underlying AI to recognize it has been asked to sit in on the same meeting, and act as the personal representative for each of 25 separate people.
If you're under the impression that there is a personal, private relationship between an individual and an AI instance, I suggest you disabuse yourself of that notion. If there is any distinction, it is only because the underlying AI has been instructed to schizophrenically simulate it.
Lol, well I can't say I'm surprised your bullishness on the matter persists - although I would argue that A) the current architectural model is a single process running the translation in a user-scoped session, and B) given that a bot literally can't recognise anything, any such implementation would be entirely conventional engineering, not waves hands AI voodoo magic. So I don't share your stance. But also I genuinely don't care, so that's as much attention as I'm prepared to give this particular thought experiment.