About unsupervised language models, or unicorns in South America

Author here. This is my first relatively serious try at writing an article about current events in the field of ML/AI, which is something I've wanted to do for some time now. All feedback is welcome, although I know attention is a scarce resource in the internet, so my expectations are adequately low :)

I tried to hit a scientific communication tone. My background in ML is relatively shallow still, although I'm working on it. It's a work in progress.

This article goes into a very positive (very likely too positive) direction, and glosses over lots of complexity and problems along the way to making good use of computational agents. My intent is to write a series of more in-depth articles on the roadblocks and dangers of this path, or similar paths, and to link to research in this field as I find it (and try to understand it).