For me the weird part is that they put in a gif of a researcher using a 'video wall' (apparently a projector combined with some gesture recognition?) to show one of these ice proteins...
Except the one shown looks like it is some helix-rich structure with a haeme group. Even if that is an ice-forming protein, I would have shown one with a beta helix instead, which have the properties mentioned - large surface for nucleation.
Read down a bit and my intuition as to why turns out to be bogus.