Comment by exitb
7 hours ago
Nothing silly was said really. The remark was specifically about local inference and RAM supply issues and it’s entirely true - if all current AI users tried to switch to local models, it would make the RAM availability situation much worse. I userstand that regular Twitter folk doesn’t grasp that, but this isn’t Twitter.
Would it? First of all, all current AI users still need memory in their devices. You can't use ChatGPT with pen and paper. There would always be a consumer demand for memory, even without local inference. Secondly, it'd drastically decrease the demand for HBM, which has shit yields when compared to conventional memory. It's also questionable as to how much of the produced HBM is actually in use and not sitting in a warehouse.
Altman is still lying by omission[0] by implicating that various important costs are not real
[0] https://news.ycombinator.com/item?id=50400515 for longer comment on that
And "if all current AI users" is a strawman, people using free chatgpt once or twice a day are not going to buy RAM to run own LLMs.
You are misunderstanding the quote.
In which way?