Ogilvie Law Firm is built on responsible use of AI. My AI Patent Guide™, Step Forward™, Check A Patent™, Check A Pursuit™, Check Both™, and Japanese translation checking legal services all include my use of proprietary AI tools I developed based on my 30+ years of practicing patent law and my 40+ years of education and experience in software generally and AI in particular, including a Master’s degree in AI. My Safe Inventing With AI™ workshop teaches inventor teams how to avoid or reduce risks when using AI to help describe and explore their inventions. My website helps people learn about these services. I also designed and implemented the website and its supporting backend infrastructure using options, scripts, schemas, settings, code, and test cases I arrived at interactively during weeks of interaction with AI chatbots.
AI Is a Tool, Not a Substitute for Human Expertise
Having access to an AI chatbot trained on chemistry and biotech does not make me an expert in chemistry or biotech. That’s one reason why I do not accept matters that would require me to understand those technical areas at a professional level. Similarly, I can use AI to better understand how a European Patent Examiner might approach an application, even though I spent years writing patent applications for Microsoft that went international, and worked directly with European counsel. But in this situation understanding alone is not enough. I am not licensed outside the United States, so I screen out matters with no US connection.
In both scenarios AI extends what I can do within my competence. But AI does not create competence I do not have. I rely on my own competence; no output from an AI tool leaves my practice without me reviewing it first. I review every draft my internal AI tools produce before a final version goes to a client.
What Irresponsible Use Looks Like
Irresponsible AI use in patent practice is easier to recognize with examples. Two patterns I have seen:
Pitfall
Asking AI to generate the invention. A prompt like “give me an invention that will help people feel better about themselves” produces a desired result without any technical mechanism being provided by the person who would be named as inventor. The AI obliges by providing output, but the human has not contributed anything technical, so what comes back is not their invention. Patents protect inventions conceived by people. If someone named as an inventor did not conceive the claimed technical solution, the resulting patent is at serious legal risk.
Pitfall
Asking AI to identify and rank alternatives. A prompt like “tell me how my invention is better than the top 100 alternatives” creates two problems at once. First, the AI will identify prior art (existing products, patents, or publications) that the inventor may then be required to disclose to the USPTO. If the AI listed a large number of references, or references that mostly overlap one another, or references that are not relevant to the patent claims ultimately filed, this AI output creates work, expense, and risk. Second, the AI’s comparative conclusions could be characterized as legal conclusions, not technical facts. Even so, they might need to be disclosed to an opponent and they could open the door to other undesirable disclosures. Relying on them can also produce a distorted picture of patentability that leads inventors to poor decisions. Responsible AI prompting in patent work is precise and bounded, and guided by measures to protect confidentiality and legal privileges. It asks AI to help describe and explore what the inventor already knows, not to draw legal conclusions or substitute for the inventor’s own technical thinking.
A Risk Framework, Not Just a Rule
Responsible AI use involves identifying risks, assessing them, and mitigating them — techniques drawn from the approach cybersecurity practitioners use to manage threats. Each of my services is designed around that framework. The AI Patent Guide™ and Step Forward™ services help inventors understand the risks that arise when they use AI to describe their inventions, and give them structured approaches that reduce those risks. The Check A Patent™, Check A Pursuit™, and Check Both™ services each applies AI-assisted analysis to identify exploitable vulnerabilities in issued patents — the kind of risks that can determine the outcome of a licensing negotiation or an infringement dispute. The Safe Inventing With AI™ workshop teaches inventor teams how to build AI workflows that protect patentability instead of inadvertently undermining it.
One overriding mitigation, in every case, is human review of AI output by someone with the experience to know what they are looking at. That human review is a practical example of the cybersecurity defense-in-depth principle.
AI Usage Is a Human Responsibility
There are social and environmental costs to AI. Like any tool, AI can be used carelessly, or maliciously. But AI is created, operated, and used by people. It is the responsibility of people to make those uses ethical, legal, and beneficial.
In that respect, AI is like flight. There are social and environmental costs to air travel, and flight technology can be used for terrible purposes. But we work together to make air travel safe, to promote its good uses, and to constrain the bad ones. Just as we seek responsible use of air travel, we can and should seek responsible use of AI. Using AI to promote and protect innovation, carefully and with appropriate human judgment at every step, is a good place to start.