@chrisstoecker@mastodon.social It's actually bad that they need multiple dozens of terabytes of RAM to get their AI to output the same quality that can be done by some random nerd's basement home lab setup using Ollama on a 2016 CPU, 128GB of RAM and a 5 year old GPU with 16GB of vRAM. Give me a break. These #AI corpo nutjobs wouldn't know good engineering if it hit them in their overbloated stock price, and EMP'd the RAM sticks that they are hoarding in their warehouses in their stupid price-fixing scheme.