Not exactly surprised here. Every time I’ve seen him on the news, it’s always him fearmongering about the dangers of generative AI, when ChatGPT is burning through money and seemed to become more and more restrictive with every iteration. You can’t run an organization if it is built on top of lies.
Actually open models (not open source, sadly) like specialized LLaMa 2 derivatives that could be ran and fine-tuned locally seems to be the future, because there seems to be a diminishing return in training/inference power to usefulness, and specialized smaller model tuned for specific applications are much more flexible than a giant general one that can only be used on somebody else’s machine.
Not exactly surprised here. Every time I’ve seen him on the news, it’s always him fearmongering about the dangers of generative AI, when ChatGPT is burning through money and seemed to become more and more restrictive with every iteration. You can’t run an organization if it is built on top of lies.
Actually open models (not open source, sadly) like specialized LLaMa 2 derivatives that could be ran and fine-tuned locally seems to be the future, because there seems to be a diminishing return in training/inference power to usefulness, and specialized smaller model tuned for specific applications are much more flexible than a giant general one that can only be used on somebody else’s machine.