LLMs are amazing, but it’s hard to watch people that give them too much credit validate their ideas with sycophant models. A safer way to use AI is to invalidate your ideas. Poke holes in them and then use your human judgment so you don’t get duped by deceptive math.

@zero_to_seed Great point. Whenever I’m testing an idea, I always ask “ what am I missing?” or “ what are the blindspots?”