Thank you for this review — always an interesting read.

My thought about searching for extraterrestrial signs with AI: it will be probably biased by human input (databases of machine learning, evaluation of new data, etc.). Probably, to make such search more efficient, they should try to minimize human-biased influence by training on recognition of unknown patterns.

Because even we humans are narrow-minded. I clearly recall that huge NASA/SETI press conference about a crucial discovery they did. Everybody was preparing to hear about some UFOs or Green Mens etc. But the news was, they found in a toxic lake in Canada living organisms. The public was disappointed (no ETs or Starships). But it was revolutionary discovery indeed— it denies our narrow conceptions of life (as urging for air, light and other, life-friendly components): in that dangerous environment, life was also possible.

We need new approaches to think bigger, and here non-supervised Deep learning can be essential for making new perspectives.