· 2 min read

AI safety ₂

How many times do I have to keep saying this.

Safety work and capabilities work are THE SAME THING. All the great capabilities advances come out of safety work! The people who actually believe in superintelligent capabilities are WORRIED ABOUT SAFETY

AND THEY ARE RIGHT

The people saying “safety is dumb, it’s just about making models woke, full steam ahead, it’s time to build”?

THEY ARE NOT CAPABILITIES RESEARCHERS

THEY ARE CLOWNS

Exactly! This is why people think “e/acc” is dumb, because the people publicly speaking for it have clearly not even done the first step in thinking about this problem, which is to ACTUALLY BELIEVE AGI IS POSSIBLE

View original