Comment by abss | Hacker News Reader

Comment by abss | Hacker News Reader

Parent Full thread

AI centered video conference application could create new revolutions in the next years. We were promised to get "AI talking heads" in SF movies 20-30 years ago. It is an obvious idea but we are not there yet. We still need a harware "revolution" that will enable systems with powerfull AI capabilities (generative AIs, capacity to train models on your data and only based on vocal indications) Imagine such environments running on desktops or even on mobile phones. This could force a migration of all software as plugins in "video conferencing" software that allow us to talk with AIs (of your employer or entertenment AIs or with other people that have AI as first filters, AI for dead people, celebrities, etc) in a safe but totally integrated approach. I am a researcher and like to imagine possible futures but defenetly there are so many opportunities for revolutions in these areas for the next 10-50 years...

This is one of those "revolutions" I struggle to wrap my head around. Even if such an AI system worked without a hitch, I think I'd end up using the traditional Skype model of video calling real people more often. It just doesn't feel "obvious" to me, but maybe I'm terribly out of the loop.

I'm not quite sure I understand the benefit of AI talking heads compared to the people you are wanting to talk to? Novelty, streamers, etc. it might make sense, but a business meeting or chat with family and friends? If someone is replacing themselves with their AI avatar then why have any sort of meeting?

It feels like the least simple solution to video calling.

If the generative part moves entirely client side, then video calls could become extremely low bandwidth. You could basically stream just the audio of the conversation, and each person could still feel like they were still talking face to face through photo realistic avatars. I don’t how likely this is, but maybe it makes reliability so good that it’s a no brainer. You could probably not even look at your camera if you’re busy, but the other person will still feel like you’re looking at them the whole time.

I don’t buy it.

I like the camera feed because I can see the person and their emotions, inflections and other mannerisms.

Some AI Borg mashup that averages every face is not going to get that for me.

Reply on news.ycombinator.com