Let’s be honest: by 2026, we’ve all started thinking about using AI to help with patent drafting. The promises of massive time-savings are just too good to ignore. But there is a massive risk to how we practice that we aren’t talking about enough. I recently came across research by Advait Sarkar from Microsoft Research, and he nails the problem. In his talk, “How to stop AI from killing your critical thinking“, he warns that we are heading toward a future of “outsourced reason”.
The Risk of the “One-Click Robot”
The problem is that many AI tools are being built as “One-Click Robots”. They jump straight into text production and automate the job so completely that we risk to stop engaging with the actual “clay” of the case.
It’s a scenario every patent attorney knows by now: You input your claims and the invention description, push a button, and get a full draft description back in minutes. The text is strikingly fluent and professionally phrased.
And here is where the trap closes. Because the draft already sounds great, it becomes incredibly difficult to meticulously go through each line and validate the strategic intent behind every sentence. We fall into “satisficing”, which is the psychological tendency to accept an AI-generated solution that merely meets a minimum threshold of correctness rather than searching for the optimal one.When we work this way, we risk becoming intellectual tourists in our own files. We “visit” the ideas by clicking a button, but we don’t actually inhabit them. Instead of being the architects of a legal strategy, we become “middle managers for our own thoughts”, as Sarkar puts it, merely validating a robot’s statistical predictions.
This creates three existential risks for the patent attorney practice:
- The “Filled Page” Problem: Moving from a blank page to an AI-filled page replaces active construction with passive evaluation. It is much harder to fix a “statistically probable” draft than to engineer a strategically sound one from the start.
- Loss of Critical Engagement: Because AI-assisted drafts look correct, they dampen our critical engagement. We risk missing the subtle strategic nuances that lead to success in the examination phase and in the courtroom.
- Mechanized Convergence: Studies show that when knowledge workers rely on “one-click” automation, they produce a smaller, more homogenous range of ideas. For us, this means patents that are more statistically “average”.
The Solution: AI-Powered “Tools for Thought”
If we want to remain indispensable, we have to move beyond “AI-assisted” automation. We need AI-PoweredPatent Drafting, where the machine acts as a Tool for Thought that elevates our thinking rather than replacing it. I am currently redesigning my claim drafting tool from the ground up to implement this exact framework (more on that soon).
It is built on three core principles:
- Material Engagement: You must “touch the clay“. The tool is designed to ensure you personally shape the protection hierarchy and define the strategic route before the machine produces a single word of claim text.
- Productive Resistance: Instead of an obedient servant, the AI acts as a provocateur. It is designed to challenge your assumptions, highlight weaknesses, and surface alternative perspectives to help you strengthen your argument.
- Scaffolding Metacognition: The tool handles the manual labor in the “data space” (the text production) so you have the headspace for situated judgment. This preserves your role as the strategic advisor who takes human accountability for the client’s case.
By following this new paradigm and leading the AI with human intent, we can prevent “mechanized convergence”, the tendency of AI to produce homogenous, statistically “average” patents.
Watch the Transition in Action
I’ve recorded a video where I share a first-look preview into my new workflow that implements the Tool for Thought paradigm.
👉 Watch: AI-Powered Patent Claim Drafting. My New Workflow.
Hope this helps,
Bastian
P.S. To move from theory to mastery, join my “Future-Proof Your Patent Drafting Workflow” seminar on 25 February 2026. We’ll get tactical: you’ll learn the workflows to automate text production so you can focus on high-value strategy. We will practice the exact prompts needed to turn AI into a tool that sharpens your judgment.
👉 Register for the Strategy Seminar here.