This highlights quite well the limits of the models used in LLMs.

Now, this is interesting research. With all that complexity, emergence is bound to happen. There's a chance to explain how and why. The links with the training data quality and the prompts themselves are interesting. It also explains a lot of the uncertainty.

Interesting take on optimisation and complexity... Indeed it's probably best to not use the latter to achieve the former otherwise that complexity can never be removed anymore.