They probably hope to run a LLM locally on the Mac mini but they don't realize that decent models require much more computational power