It's so interesting to me that people are still genuinely calling language models a "scam" in $current_year. I get useful inference out of a model running on a

It's so interesting to me that people are still genuinely calling language models a "scam" in $current_year. I get useful inference out of a model running on a rack in my living room. Where is the scam?