An interesting observation I've noticed recently is how some people trying new AI models for the first time, particularly video and image generators, often hit immediate frustration because they expect them to work like search engines or chatbots. They assume you type three words and get exactly what's in your head. But these tools have their own internal logic, their own quirks and failure modes. Like any creative medium, they require extensive experimentation to understand what they can and can't do, which prompts actually work versus which sound like they should work. You have to develop an intuition for their specific constraints and capabilities.