@suricrasia one thing i love about this is how comprehensible lethargic appears at time and how much some of these interactions feel like an actual conversation. it's nowhere near as convincing as an LLM, but you can see how the gap between the two is smaller than one might imagine, but unlike and LLM it's trivial to demonstrate that the algorithm isn't intelligent at all. it's a fantastic example of how nonsense input can be refined into something that appears to have intelligence behind it