OpenAI CEO Sam Altman said in a podcast interview that "it's quite terrible" for people to have RAM in their own PCs when it could go into AI data centers.
  • aanes_appreciator [he/him, comrade/them]@hexbear.netEnglish
    7·
    6 days ago

    I know his whole business runs on deferring computing to the cloud but is he really this dense?

    What happens when people discover you only need ~1GB to run models capable of 90% of what you’d otherwise use a cloud LLM for?

    What happens when every IT enterprise ever discovers they can save 90% of their cloud costs (and all the data compliance, availability, and reliability issues that come with it) by spending like £5k on a system that can, again, do 90% of what Altman is trying to sell you?

    What happens when people realise some of the stuff they “offloaded” to LLMs is faster, easier, and more robust when integrated into a system designed to do it properly?

    OpenAI’s hype assumed that human knowledge workers are the only workers involved in the production of surplus value. For all that surplus value extracted by laying off workers en masse, the reality has set in that LLMs are simply highlighting their own inefficiencies, and going the small hop further of addressing those makes the LLMs obsolete.