Comments by "Paul Aiello" (@paul1979uk2000) on "Is AI really getting dumber? Llama2 vs GPT-4" video.

  1. It will be interesting to see if LLAMA 2 goes through the same process that LLAMA 1 did, I remember that in the early days, LLAMA 1 wasn't that great, but over a matter of months, the quality went up a lot, as well as the performance, if LLAMA 2 goes through the same process, whiles LLAMA 2 is already starting off from a better position, the quality of LLAMA 2 by the end of the year could be crazy good. I think over the coming years, these open source A.I. are going to be a major threat to the online models, the locally run open source models have a few major advantages over the online ones, privacy, security, far fewer restrictions or no restrictions if you want that, and the performance is already at a good stage where you can run upto 30b on a cpu or gpu and 60+ billion if you have enough system memory but it's a bit on the slow side but still useable. It also looks like Qualcomm are pushing it's next gen chips with A.I. in mind at a local level, probably because many of the phone companies want to have a locally run version on their phones, and I suspect most consumers, businesses and governments would on the phone or PC. There are too many advantages to the locally run models, and I suspect it's going to become more of an issue for the online ones as A.I. becomes more useful and integrated in our lives, basically, government regulators, the public and even many companies are likely going to ramp up the pressure on the online models when it comes to privacy and security and honestly, I don't think they can win this one whiles the open source A.I. doesn't have any of these issues. It wouldn't surprise me if Microsoft and Meta already knows this and are edging their bets with closed and open source A.I. they probably realise that public, media and government pressure could build to such a level against the online models that it's not worth them doing any more, and there is also the factor that over the long run, A.I. is going to be very important to the economy and our lives, that basically means it's a high risk thing for so few to control it and again, open source has none of these issues. On the other side of the coin, open source does have it's risk in having unrestricted A.I. but I think it's better to go that route with more transparency over so few controlling A.I. with how important it's likely to be for us all.
    2