Comment by mjr00
3 days ago
> When AI systems make mistakes, it doesn’t make sense to assign blame the way we would with human errors - they’re tools operating within their programming constraints, not autonomous agents making conscious choices.
It's not really "assigning blame", it's more like "acknowledging limitations of the tools."
Giving an LLM or "agent" access to your production servers or database is unwise, to say the least.
In this thread the person does literally assign blame, accuses the AI of lying, and makes it write an apology letter to the team as though it's a child that needs to be chastised.