How AI Safety Workers Are Talking Past Each Other
AI safety experts from different worlds keep clashing — and it could slow down saving the planet
Inside co-working spaces dedicated to preventing AI from going rogue, two tribes coexist uneasily. One side has spent years studying AI safety from first principles, poring over dense theory and insider jargon. The other side brings deep real-world experience in government or industry and now wants to pivot toward AI governance.
In theory, mixing these groups should accelerate progress. In practice, they speak different emotional languages. One group expresses existential dread about AI wiping out humanity; the other shares inspiring stories about talented teams. These reactions confuse each other, creating awkward moments that push people apart instead of pulling them together. Meanwhile, policy discussions stall on fringe ideas that feel absurd to one side but serious to the other.
The real risk isn’t just awkwardness — it’s delay. If these two cultures can’t find a way to collaborate, the chance to steer AI safely could slip away. Someone needs to step in and help them translate their fears, goals, and jargon. Until then, the clock is ticking.
- Two groups working on AI safety — theorists and experienced professionals — barely understand each other
- They react emotionally in ways that confuse the other side, making collaboration harder
- Without better communication, progress on preventing AI disasters could slow down
Why It Matters
Faster AI safety progress needs these groups to understand each other — before time runs out