Alice Step 2 Approach for AI Inventions at the USPTO

Conversely, abstract ideas can be dressed up with specialized machine learning accelerators to wrongly appear eligible, despite being ineligible.

As a quick reminder, deciding patent eligibility under the Supreme Court’s 2014 Alice Corp. v. CLS Bank International involves two steps. Alice Step 1 checks if a claimed invention is directed to an ineligible concept, like an abstract idea. If it is, Alice Step 2 looks for an «inventive concept» to ensure the claim isn’t just a way to monopolize the abstract idea.

Recently, Alice Step 2 at the USPTO has devolved into merely checking for the use of a generic or conventional computer. This is a flawed test, as it’s over-inclusive and often a foregone conclusion for computer-related inventions. Consequently, AI-related claims failing Alice Step 1 rarely get a fair second chance at Step 2.

Over three years without the PTAB finding a claim that satisfies Alice Step 2 highlights this issue. This shift towards a predetermined Alice Step 2 seemingly responds to the burdens imposed by the Berkheimer Memo, which required extensive fact-finding for examiners.

As a matter of fact, the USPTO is revising its eligibility guidance for AI inventions. It should clarify that claims should be found ineligible under Alice Step 2 not just for reciting generic or conventional computers, but for including only features inherent to monopolizing an abstract idea. Restoring Alice Step 2 is vital to ensure sufficient protection for significant AI inventions.