People love arguing about how to make AI "safe" (using this term very loosely here) because "safety" is something we value in the same way that in prior centuries we valued religious piety and because AI of the types we're arguing about isn't really implemented in the real world yet we can spout whatever garbage we want and nobody is going to reply with a "well akshually" that cites reality. Basically the discussion makes for a great circle jerk among technical folk. And as others have mentioned, progress marches on regardless.