AI Guardrails Issue

AI guardrails are needed to prevent use of historical information for discrimination, hate, etc - also for other topics besides history. 

But guardrails may unintentionally introduce asymmetric contextualization that sometimes sounds like there is an intentional "sanitizing" of history - historical events framed differently depending on the groups involved or the recency of the conflict. 

Some even point that the guardrails design is following the mantra "the end justify the means" - or "the end justify the means because you can’t make an omelette without breaking eggs". 

The important thing is that AI is a tool, not an oracle. It helps as the starting point in research. We need to do the homework ourselves.

Interesting the answer from one AI model to one of my prompts.

The irony is that AI was sold as the "ultimate library"—a way to access all human knowledge instantly. Instead, it is becoming a "Librarian with an Agenda.

Looks like the system is aware of the problem.