Agree - AI writing is not the same as writing with AI :-)

That’s a solid starting point, Sabrina, especially for short tasks or early exploration. The problem appears when people try to move from simple questions to real professional work, such as proposals, business cases, RFPs, or long-form documents. Before the actual work even begins, a significant portion of the context window is consumed by explanations, clarifications, and analysis. As the conversation grows, responses slow down, logic starts to drift, previously confirmed decisions quietly change, and parts of the text are compressed or disappear without warning. Staying in the same chat makes this worse over time. Starting a new chat helps only marginally. Not all context carries over, and if the original conversation is already fragmented, the new one inherits that fragmentation. This is why even a very well-crafted prompt eventually fails. Prompts are useful, but they are not a reliability mechanism. What does work is understanding these limitations and adding discipline and governance to how we collaborate with AI. Without that, AI may feel helpful at first, but it becomes unreliable for sustained, professional work. There is a big difference between letting AI write for you and writing with AI in a controlled, deliberate way.

Also, it helps if you explain to AI what you actually want to achieve and what information you currently help.

Have several articles about it on LinkedIn