It is no longer a question of whether lawyers are using AI. They are. Adoption has moved quickly from curiosity to everyday workflow, and many lawyers are now turning to these tools not just for editing or summarizing, but to help generate arguments, outline positions, and frame legal analysis. In some cases, the first draft of a brief or email is no longer coming from the lawyer. It is coming from a prompt.

In the past, when a lawyer received a new matter with complex legal issues, the lawyer would review the facts, identify the legal questions, and begin pulling relevant authority. There might be some time spent sitting with the problem, thinking through different angles, weighing possible arguments, and considering alternative approaches before anything was written. That early stage was not wasted time. It was where the real thinking happened. It was where the lawyer organized her thoughts.

Now, that step is increasingly being skipped. Instead, the lawyer uploads the issue into an AI tool and asks for an analysis, an argument, or even a complete draft, relying on the output as the starting point rather than their own reasoning.

It feels efficient. It feels modern. It feels like you are getting a head start. But there is a subtle shift happening in that moment, and it may not be a good one.

Lawyers have traditionally been trained to think first and write second. The act of working through a problem is where the real value lies. Writing is not just the output. It is part of the thinking process. It is how you discover what you actually believe.

The “prompt first” approach flips that sequence. It invites you to outsource the initial thinking and start with a fully formed answer. The AI gives you structure, language, and conclusions before you have done the hard work of analysis yourself. That can feel like momentum, but it skips the most important part of the job.

When you start with your own thinking, you are actively engaging with the problem. You notice gaps. You ask better questions. You spot the issue that is not obvious on the surface. When you start with a generated answer, you are more likely to react to what is in front of you. You edit. You tweak. You polish. But you may never fully interrogate whether the underlying reasoning is sound. That is how hallucinated AI results slip through the cracks.

We’re not only talking about simple incorrect citations or misstatements of law. The more concerning mistakes are the subtle ones. There are arguments that are technically correct but strategically weak or issues that are framed too narrowly. Those are harder to catch because they look right.

A lawyer’s duty of competence includes the obligation to exercise independent judgment. That does not mean you cannot use tools. It means you cannot rely on them in a way that replaces your own analysis. If you start with the prompt and skip the thinking, you are moving in that direction.

And in the long-term, this cognitive offloading can impair your ability to analyze and think critically. Studies show a strong negative correlation between AI usage and critical thinking.

The good news is that this problem is avoidable. Using AI does not require you to surrender the part of the process where legal judgment is formed. It simply requires some discipline about when the tool enters the picture. The goal is not to avoid AI altogether, but to make sure it comes after your own initial analysis rather than before it. In other words, AI should supplement your thinking, not substitute for it.

Pause before you prompt. Take a few minutes to outline the problem in your own words. Identify the likely issues. Think through possible outcomes. Even a rough mental sketch is enough. Then, if you use AI, use it as a second step. Let it challenge your thinking, not supply it.

When you review the output, do not just edit for style. Interrogate it. Ask what is missing. Check whether the conclusions actually follow from the facts. Verify the law. Treat it as a draft that needs real scrutiny, not a finished product that just needs polishing. Don’t be afraid to enter a prompt that asks, “Are you sure that is right?” In the end, the real value a lawyer brings is not the ability to generate an answer quickly. It is the ability to exercise judgment, spot what matters, and think through a problem in a way that serves the client’s interests. AI can be a useful tool in that process, but it is a poor substitute for the disciplined reasoning clients are paying for. If we are not careful, the convenience of prompting first can quietly weaken the very skill that defines good lawyering. That is why the order matters. Think first. Prompt second. And keep the lawyer, not the machine, in charge of the analysis.