Clearly BackChannel17 doesn't understand the Nothing Burger reference (especially given I've been eagerly spreading it myself).
Nothing Burger was a verbatim quote recorded by Project Veritas (great site, excellent journalism - what media outlets should be doing) when they caught New York Times bragging that the whole Russian accusations against Trump were a big "Nothing Burger".
So when BC17 tries to claim their own words are 'not a "nothing-burger"' (as though accused as having been), they're basically insinuating they're batting for the false media IE the New York Times.
Questionable indeed.
The only element I can confirm is the quotations of "Hyperparameter Tuning Capabilities", which can be found cited here:
https://arxiv.org/pdf/1711.05869.pdf
It's a bit of a boring read to be honest, and I think it talks about predictive neural networks, but it's so dull and uninteresting (and I've read some pretty bad studies before) I'm not inclined to try to sift through it for gems.
I did a quick eyeball on the two named individuals. They just look like University researchers, no obvious CIA connections.
This isn't the most damning study on arxiv. Try looking into topological control of social networks (IE manipulation of how people behave on social networks). Censorship AI is pretty old hat. Adaptive censorship AI is very old hat.
PS: You can trick adaptive censorship AI to censor the 'wrong' things. They also seem pretty bad at distinguishing images with a lot of noise in them.
But I digress.