I am curious about Luxmain’s multiple points of view on Gaming and how AI is use
Just ask away if you have a specific question. I have several decades background in each of AI and gaming so I can’t really just dump it all in a post.
I know a little that GPUs now use some AI. That is different than many would expect.
It’s more accurate to say that the driver software that makes GPUs run, can use AI techniques to accelerate performance. I wonder if I can give an analogy. It’s a bit like, rather than actually do all the computational work to draw, say, a butterfly from a 3d model and some textures and colourings, the GPU driver now sends a command ‘draw a butterfly’ and a statistics library just drops points down that are butterfly-ish, statistically, in that area.
Some AI techniques use GPU hardware.
Some GPU software uses AI techniques with GPU hardware, to quickly do things that would previously have required lots more GPU hardware (and energy).
I know moderation can be done with AI.
I would look up the topic of sentiment analysis. Moderation has been done with ai ever since someone created a ‘naughty words’ list to filter. The question is, how complex is the processing of the words and sentences? Sentiment analysis is about figuring emotional intent of some words. Are the words intended to hurt? Do they express happiness or hate? etc
Generally speaking most moderation is plain old text matching. You could do something fancier. Ultimately though, a bad person will take the moderation software and ‘test’ what they’re going to write, to ensure it will sneak through moderation. No matter how you write the software.
I’m sorry, I don’t really have time to listen to podcasts or watch youtubes at the moment. If you can give me a 2-3 sentence summary, happy to give feedback.
Zuckerberg and one of his IT managers(?) showing off a newer mapping system for facial stuff, movements, speaking, and emotions showing…The quality is right down to blemishes on the skin. It is not cartoonish.
The next phase according to Zuck is to reduce to one photo the needed data to make these AR/VR communications.
Facebook have IMHO probably the strongest group of AI projects other than say NVIDIA, among ‘big AI companies’. And not just recently, for a long time. Last year, their efforts in ‘metaverse’ were being mocked for cartoonish stupidity. This new approach to avatars adds credibility to their metaverse project, demonstrates capability in AI, and may be successful in creating a novel VR or AR based system of communication for business.
I think also the idea of an avatar of yourself you can ‘adjust’, and bring with you into different games has a lot of potential.
I don’t hold META but it might be the least bad overall among the AI tech companies in terms of stock price, generation of potential new avenues of income etc.
Reducing to a single photo may or may not work well, but, there are already techniques that take 3-4 ‘random different angle’ shots and combine them into an excellent overall result. The output looks better than many 360’ photogrammetry approaches.