Bravo
If ARM was an arm, BRN would be its biceps💪!
Aiieeeyyyee!
On this Youtube interview from 3 days ago, Sam Altman is asked a question about the prospect of developing an on-device model for latency. His response is very interesting and it seems IMO like he believes there will be a necessity for it, especially for generated video content with AR goggles in real-time.
(3.50 mins)
Interviewer :"Do you think you need to develop an on-device model to decrease latency to the point for usability."
Sam Atman: "Umm for video, maybe it would be hard to deal with network latency at some point, like, the thing that I've thought would be super amasing would be to put on someday a pair of AR goggles or whatever and just speak the world in real-time and watch things change and that might get harder over network latency. But for this (meaning ChatGPT 4), 200 or 300 milliseconds of latency feels super, like, it feels faster than a a human responding to me in many cases".
Interviewer "Is video in this case images?
Sam Atman: "Oh, sorry I meant video if you wanted generated video, not input video".
Interviewer " Got it. So currently it's working with actual video the way it is.
Sam Altman: "Well, frame by frame".
On this Youtube interview from 3 days ago, Sam Altman is asked a question about the prospect of developing an on-device model for latency. His response is very interesting and it seems IMO like he believes there will be a necessity for it, especially for generated video content with AR goggles in real-time.
(3.50 mins)
Interviewer :"Do you think you need to develop an on-device model to decrease latency to the point for usability."
Sam Atman: "Umm for video, maybe it would be hard to deal with network latency at some point, like, the thing that I've thought would be super amasing would be to put on someday a pair of AR goggles or whatever and just speak the world in real-time and watch things change and that might get harder over network latency. But for this (meaning ChatGPT 4), 200 or 300 milliseconds of latency feels super, like, it feels faster than a a human responding to me in many cases".
Interviewer "Is video in this case images?
Sam Atman: "Oh, sorry I meant video if you wanted generated video, not input video".
Interviewer " Got it. So currently it's working with actual video the way it is.
Sam Altman: "Well, frame by frame".
Last edited: