Even then, AI models (be it text or image) are generally unethically trained (i.e. without consent of the authors/artists of the training material) and have a significant energy consumption, even for single prompts.

And I do have to ask: To what degree is running your comment through an LLM actually beneficial? You say it improved readability, but how unreadable was your original comment actually, that it would require fixing via external tool?

Ah, that‘d make sense

I never quite got how interviews could be rare. Lost, maybe. But as soon as it’s on YouTube, how can it be rare? It’s right there, available to everyone, any time.