Agreed. Also, what exact architecture were they using, how were they segmenting, etc. This feels like a rehashing of known challenges using a random implementat

Agreed. Also, what exact architecture were they using, how were they segmenting, etc. This feels like a rehashing of known challenges using a random implementation struggle to make some broad "AI" judgement.