Ёлки-палки, плохи наши дела ))

LLMs can function not only as databases, but also as dynamic, end-user programmable neural computers.

LLMs cannot reason. Many are working on fixing this. For example, below is a proposal to add logic programming (Prolog) by teaching an LLM to translate natural language queries into Prolog statements, which are then answered by a Prolog processor...
... the integration of Prolog enables LLMs to achieve high performance on the NLR dataset, which even the most advanced language models (including GPT4) fail to solve using text only.

Автор тот самый Эрик Мейер ("Why the fuck do we let people that have no computer science backgrounds write code? Unacceptable!").