That's definitely it.

This probably means that OpenAI et al are fine-tuning salesman-like LLMs to "fix" that problem.

Can't wait for the future.