[RandomLensman]

No, most/a lot of activities are low risk, but my point is we seem to struggle with AI in high risk while we automate human flaws.

Also, LLMs don't understand other than via the language representation.