avturchin comments on AI safety via market making