Key Changes:
- The «Human Mind» Test Gets Refined
One of the most significant clarifications involves the mental process analysis. The USPTO emphasizes that claims are only abstract ideas if they «can be performed in the human mind, or by a human using a pen and paper.»
AI-related claims that cannot practically be performed by humans don’t automatically fall into the abstract idea category. This could open doors for complex machine learning algorithms that were previously rejected.
- «Recites» vs «Involves» — A Critical Distinction
The memo draws a sharp line between claims that actually describe an abstract idea versus those that merely use one. This distinction could save many AI patents from unnecessary rejections.
Example from the guidance: A claim limitation about «training the neural network in a first stage» doesn’t recite an abstract idea, even though training involves mathematical concepts. However, specifically calling out «backpropagation algorithm and gradient descent algorithm» does recite mathematical concepts.
- Holistic Claim Analysis Required
Examiners are now explicitly instructed to analyze claims as a whole rather than picking apart individual elements. The memo states that «additional limitations should not be evaluated in a vacuum, completely separate from the recited judicial exception.»
This holistic approach could benefit complex AI systems where individual components might seem abstract but work together to solve specific technological problems.
- The 50% Rule for Rejections
Perhaps most importantly for applicants, the memo establishes that examiners should only reject claims when it’s «more likely than not (i.e., more than 50%)» that they’re ineligible. Uncertainty alone is no longer grounds for rejection.
- Technology Improvement Remains Key
Claims that show genuine technological improvements are more likely to be patent-eligible. The guidance emphasizes providing «particular solutions to problems» rather than merely automating existing manual processes.
Practical Tips:
For Applicants:
— Emphasize technological improvements in your specifications and claims
— Describe specific solutions rather than general outcomes
— Highlight aspects that cannot be performed manually by humans
— Show how your AI system solves technical problems rather than just business problems
For Attorneys:
— Use the «recites vs involves» distinction in arguments against Section 101 rejections
— Point to the 50% standard when examiners seem uncertain about eligibility
— Emphasize whole-claim analysis rather than allowing piecemeal evaluation
Bottom Line:
The new guidance signals a more patent-friendly approach for genuine AI innovations, but inventors must still demonstrate real technological advances—not just automation. For companies investing in AI R&D, this provides a clearer roadmap: focus on technological improvements, describe specific solutions, and innovate rather than simply automate.
