An LLM is an ordered series of parameterized / weighted nodes which are fed a bunch of tokens, and millions of calculations later result generates the next token to append and repeat the process. It's like turning a handle on some complex Babbage-esque machine. LLMs use a tiny bit of randomness ("temperature") when choosing the next token so the responses are not identical each time.
But it is not thinking. Not even remotely so. It's a simulacrum. If you want to see this, run ollama with the temperature set to 0 e.g.
ollama run gemma3:4b
>>> /set parameter temperature 0
>>> what is a leaf
You will get the same answer every single time.
The ribbon was contentious but most people are familiar with it and it has advantages like taskcentricity and less clutter. LibreOffice has an experimental ribbon that I think should be worked on, mainstreamed and set during installation or in the settings.
UX in other areas should be improved. Lots of little annoyances add up for new users and can break their opinions. It's not hard to look over the UI and see things which have no business being there, or should only appear in certain contexts, or could be implemented in better ways. I think the project should get some MS Office volunteers into a lab and ask them to do things and observe their problems. I'd have power Word, Excel, Powerpoint users come in and do non-trivial things they normally do and see where they trip up or even if they can do what they need.