This is repeating the history of AI research "safety": 1. Let us make sure our research makes the world better, not worse. 2. That's hard. 3. Instead, focus

This is repeating the history of AI research "safety": 1. Let us make sure our research makes the world better, not worse. 2. That's hard. 3. Instead, focus on race and gender as approved by Twitter. 4. Call (3) "safety". 5. Researchers previously focused on (1) rebrand to (3). See the transition from (1) "safety/alignment" to (2) making sure that the prompt "doctor" generates images of women instead of just men. People have a general sense that some science can cause harm. (See, the atom bomb.) Instead of thinking hard to prevent that, they run to "don't make Twitter angry about gender/race."