I'm not sure the authors realize this is way more than "just a cool demo": if this holds up, it's going to be huge for game QA work.
Instrument your game to output properties of entities near the player and the output is the various control inputs - moment to moment gameplay gets solved. Maybe augment with a tick-by-tick controlled stepping mode if particularly twitchy - an LLM can take care of the higher level reasoning then.
The doom video is also in the article itself (headline: "Doom").
I suppose this is the same video as the one from the parent comment, but I don't know for sure - I don't have a twitter account and the above link doesn't work for me.
But when their system is given the instruction "do not fire, simply dodge" - it doesn't "simply dodge", it actually gets close to the fleshy pink demon rather than keeping its distance. Or am I misunderstanding?
The doom demo is quite cool