gcm

Language emerged from cognition, so it's hard for me to imagine the inverse. However, Thinkish and Neuralese do make that leap more legible in my mind. 

Also, is this too brittle a solution: hook on a classifier model to filter out anything that drifts from the soul document?

Thanks for sharing. 

Agreed, the risk of mirror-life should be taken seriously even if its viability is far off into the future.

Here's a fun thought experiment: https://archive.ph/t1Ulh#30%^[1]

It's a hypothetical scenario where we engineer a mirrored organism (e.g. maybe cyanobacteria?) to produce L-glucose to sell as artificial sweetener. It ends with the Earth becoming an ice-rock.

1. ^{^}

   It's an archived link because the author has since deleted the post upon request of a biosecurity expert at OpenPhilanthropy