I'm putting my money on CryoEM solving all the relevant structural problems first. Plus, I like exotic electron optics and UHV much more than ML - at least you get some expensive paperweights when a turbopump grenades...