A topic about AI that is not discussed / researched enough is the mental drain of learning new concepts with LLMs. This week I have spent much free time experimenting with writing a game in Unreal Engine. I‘ve never done something like this, have no clue about 3D games, and I use an LLM to teach me what to do. So in this respect I‘m probably like junior developers coming out of university.

And while I‘m FAST with LLM help, getting things done much faster, the „new concepts to understand per time unit“ ratio is high. Too high. After two hours or so I am exhausted, trying to digest all that the LLM is throwing at me. Without AI help, I would probably be still in my baby steps, while in reality I have created a first room, different animations for my character, an overlay with progress bars, and C++ code to compute what these overlays show.

I believe this is a good way to learn, because your LLM teacher will answer questions until you understand what‘s going on. But if I imagine young people right out of university getting bombarded with concepts like this 8 hours a day, I can‘t help but feel sad for them. Because unlike me they can‘t turn off their computer and say „okay, 2 hours, I‘m done for today“.

(In case you‘re interested: I‘ve come across less hallucinations than the error rate I would expect from a human teacher; LLM was GPT5.2)

#llm #learningWithAI #mentalLoad #mentalDrain