ParentFull threaddkjaudyeqooe·We need an inverse GPT4-style LLM that doesn't provide answers but instead asks relevant questions.View on HN