I think it would be very helpful to do an explicit cost effectiveness model for approval voting. Of course there would be a lot of uncertainty, but not necessarily more than AI or alternate foods. It could be for the present generation (like this or this for alternate foods) or for the long term future, like this for alternate foods and AI. Then we would have at least some quantitative way of comparing.
I think it would be very helpful to do an explicit cost effectiveness model for approval voting. Of course there would be a lot of uncertainty, but not necessarily more than AI or alternate foods. It could be for the present generation (like this or this for alternate foods) or for the long term future, like this for alternate foods and AI. Then we would have at least some quantitative way of comparing.