the fact that it is theft
There are LLMs trained using fully open datasets that do not contain proprietary material... (CommonCorpus dataset, OLMo)
the fact that it is environmentally harmful
There are LLMs trained with minimal power (typically the same ones as above as these projects cannot afford as much resources), and local LLMs use signiciantly less power than a toaster or microwave...
the fact that it cuts back on critical, active thought
This is a usecase problem. LLMs aren't suitable for critical thinking or decision making tasks, so if it's cutting back on your "critical, active thought" you're just using it wrong anyway...
The OOP genuinely doesn't know what they're talking about and are just reacting to sensationalized rage bait on the internet lmao