Think about what the AI set out to do:
- Extract and sharpen the inventive concepts (what’s actually patentable here).
- Run a §101/102/103/112-aware patentability analysis against Transformers, MoE, multi-tower, CBM, modular DL, etc.
- Propose a claiming strategy + independent claim skeletons (US-centric, EPO-aware).
- Flag edge cases / assumptions where we may want a second filing or CIP.
Most patent attorney have trouble with at least one of these areas.
Step 1: Just to be able to grasp the inventive concepts is a big deal that many get lost at. It is common for them to confuse the the embodiment with the invention or to struggle to understand how the invention is seen from the point of view of a person skilled in the field. The law requires that a patent attorney be also competent in the technical field. So most patent attorneys are also engineers. But the law doesn't really say how competent they must be. It is easy for anyone to not be able to keep up with the developments in an technical field.
So step 2, which is the core invention analysis, is dependent on step 1. You'd better hope that the paralegal did a decent job mapping your invention to the prior arts and that your lawyer was able to see the nuances.
Step 3, is the next core competency for a patent lawyer. But again, it is not as easy. There are many cases that negate each other. And as I said, the law is often intentionally vague and ever changing. Importantly, as I showed, the AI was not only able to discern that there were 3 inventions hidden in my white paper, it mapped a strategy that considered previous filings and how this should fit within the portfolio and even looked ahead to strategize for future filings (both internationally and as extension to the current one).
Step 4, is often problematic for most people in all fields. One the whole, people have a hard time knowing all their assumptions and feeling how realistic their worldview is and what their own bloodspots are. An AI has an easier time here because you can iterate between writing a patent and attacking it as a prior art you want to defeat.
I may roll this out as an app/service at some point. But the main reason for that post was to show how well the AI is able to think and infer the intent and what to do even when the subject matter is inherently vague and difficult.
I can switch some training and have the exact same architecture reason in the medical field. Or be a therapist. |
