Introduction
Is the claim as a whole directed to an abstract idea? Based on the Alice decision, that is the central question that courts ask themselves in determining whether a claim is patent-eligible under 35 U.S.C. § 101.
As indicated in Alice, the Supreme Court's primary concern was that claims covering an abstract idea “would pre-empt use ... in all fields, and would effectively grant a monopoly over [the] abstract idea.“1
The Court sought to prevent patents covering the “building blocks” from tying up all ways of implementing those building blocks, in favor of patents that integrate the building blocks into a meaningful and targeted application of the abstract idea.
The foundation of the Court's analysis in Alice was the identification of the abstract idea, and the abstract idea was then used as the basis for determining whether the claim included “additional features” to ensure that the claim did not monopolize the abstract idea.
This “outside-in” approach focuses on the claim as whole first, and then analyzes particular claim limitations if needed, consistent with the role of § 101 being a “coarse … filter” for patent-eligibility.2
Yet, the latest USPTO Guidance regarding subject matter eligibility of AI-related inventions departs from the approach outlined in Alice. Instead of identifying an abstract idea that represents the claim as a whole, the examples in the new AI Guidance dive headfirst into a limitation-by-limitation analysis to determine whether the claim is eligible.
In other words, the Guidance analyzes claims from the inside out, rather from the outside in, as required by Alice and other Supreme Court precedent. Moreover, by not grounding the limitation-by-limitation analysis to any particular benchmark, the new Guidance will likely cause unnecessary confusion and delays for patent examiners and patent practitioners alike.
And when faced with unclear or confusing guidance, the examining corps' natural reaction is to maintain rejections, while awaiting clearer guidance or case law that paves a path toward overcoming the rejections.
The Supreme Court Has Applied an Outside-In Approach When Evaluating Section 101
Under § 101, one is entitled to a patent for any new and useful process, machine, manufacture, composition of matter, or any “new and useful improvement thereof.” Courts have created numerous judicial exceptions to this general rule, which exclude from patentability any invention that is directed to a law of nature, a natural phenomenon, or, relevant here, an abstract idea.
The Supreme Court's decision in Diamond v. Diehr laid the groundwork for how patent examiners should evaluate claims under § 101.3 Importantly, the Court held that claims should not be dissected into “old and new” elements, and instead must be evaluated as a whole in determining patent eligibility.
Even though the claim in Diehr employed a well-known mathematical equation, the Court held that the claim as a whole would not preempt use of that equation and was therefore directed to statutory subject matter. The Court's decision in Diehr provided a predictable and objective test for evaluating claims as a whole, not in piecemeal.
As the use of computer software became more ubiquitous, the section 101 analysis became more complicated. In Alice, for example, the Court held that claims directed to mitigating settlement risk involved an abstract idea because they involved a “fundamental economic practice.“4
The Court was concerned that patenting such abstract ideas would “effectively grant a monopoly” over that idea, thereby “imped[ing] innovation more than it would tend to promote it.“5 Like the decision in Diehr, the Court in Alice looked first at the claim as a whole and determined that the claim was directed to a fundamental economic practice.
Only after that initial determination did the Court look to the specific claim elements themselves to see if they recited “something more.” It reasoned that a claim reciting additional features did not present the same preemption concerns because the claimed improvement would not “disproportionately t[ie] up the use of the underlying ideas.“6
The outside-in approach to the § 101 analysis is crystalized in the Alice and Diehr decisions, suggesting that any proper § 101 analysis should first determine whether the claim as a whole is directed to an abstract idea, and, if so, what that abstract idea is.
From there, the claim elements themselves would then be evaluated to determine whether the claim recites something more than the abstract idea. Indeed, courts have largely followed this outside-in approach in the years since the Alice decision.
For example, in the BASCOM 7 case, the Federal Circuit analyzed claims directed to filtering internet content. The panel held that while the claims were directed to the abstract idea of filtering content at a high level, the claims did not “preempt all ways of filtering content on the Internet.”
Rather, the claims recited a “specific, discrete implementation of the abstract idea of filtering content” and thus contained an inventive concept amounting to “significantly more” than the abstract idea. Importantly, the Federal Circuit's analysis focused on explicitly identifying the abstract idea — here, filtering internet content — prior to conducting the remainder of the Alice analysis.
USPTO's AI Guidance Analyzes Claims from the Inside-Out
While the Supreme Court took an “outside-in” approach that begins with the claim as whole, the USPTO takes an “inside-out” approach that begins with individual claim elements.
In particular, as stated in the MPEP and the July 2024 Guidance Update on Subject Matter Eligibility, an Examiner should “identify[] the specific limitation(s) in the claim under examination that the examiner believes recites an abstract idea, and … determin[e] whether the identified limitations(s) fall within at least one of the groupings of abstract ideas.”
As shown in the Subject Matter Eligibility examples, this “inside-out” approach amounts to parsing a claim into sections and identifying purported abstract idea(s) recited in each section.
And contrary to the Alice decision, the Guidance never attempts to identify what the claim as a whole is directed to, whether that claim would cover an abstract idea that would preempt use of the abstract idea, and whether the claim included “additional features” to ensure that the claim did not monopolize the abstract idea.
But purportedly identifying abstract idea(s) in this manner yields unintuitive results that are incompatible with judicial precedent.
For example, claims that expressly recite mathematical operations may be found patent ineligible by examiners due to particular limitations, whereas broader claims that recite the result of such mathematical operations may be patent eligible.
Such a result is incompatible with the reasoning of the Supreme Court in Diehr that “a claim drawn to subject matter otherwise statutory does not become nonstatutory simply because it uses a mathematical formula, computer program, or digital computer.”
Furthermore, contrary to Federal Circuit precedent and the USPTO's Ex Parte Hannun decision 8 (designated as Informative), the “inside-out” approach may find that a claim with limitations that cannot practically be performed in the human mind recites a mental process.
Example 48 in the AI Subject Matter Eligibility Guidance demonstrates the unintuitive and incompatible results of the “inside-out” approach.
This example concerns an artificial intelligence-based method of analyzing speech signals and separating desired speech from extraneous or background speech. Example 48 includes, in part, a patent-ineligible independent claim 1 and a patent-eligible dependent claim 2:
1. A speech separation method comprising:
(a) receiving a mixed speech signal x comprising speech from multiple different sources sn, where n ∈ {1, ... N};
(b) converting the mixed speech signal x into a spectrogram in a time-frequency domain using a short time Fourier transform and obtaining feature representation X, wherein X corresponds to the spectrogram of the mixed speech signal x and temporal features extracted from the mixed speech signal x; and
(c) using a deep neural network (DNN) to determine embedding vectors V using the formula V = fθ(X), where fθ(X) is a global function of the mixed speech signal x.
2. The speech separation method of claim 1 further comprising:
(d) partitioning the embedding vectors V into clusters corresponding to the different sources sn;
(e) applying binary masks to the clusters to create masked clusters;
(f) synthesizing speech waveforms from the masked clusters, wherein each speech waveform corresponds to a different source sn;
(g) combining the speech waveforms to generate a mixed speech signal x' by stitching together the speech waveforms corresponding to the different sources sn, excluding the speech waveform from a target source ss such that the mixed speech signal x' includes speech waveforms from the different sources sn and excludes the speech waveform from the target source ss; and
(h) transmitting the mixed speech signal x' for storage to a remote location.
Example 48 describes the patent-eligibility analysis for each of these claims. Using the “inside out” approach, the Guidance proceeds limitation-by-limitation to indicate whether they recite mathematical operations and mental processes.
Mathematical operations
Example 48 identifies steps (b) and (c) of claim 1 as reciting mathematical operations and thus patent-ineligible. But the analysis of Example 48 is inconsistent with Supreme Court precedent and would lead to an illogical result.
For example, under the reasoning in the AI Guidance, removing the particular mathematical operations recited in (b) and (c) — i.e., broadening the claim — would yield a patent-eligible claim under the “inside-out” approach of the AI Guidance. In fact, broadening the claim language would indisputably yield a patent that has a greater preemptive effect than a narrower claim directed to particular mathematical operations.
Of course, claim 1 of Example 48 could be found patent-ineligible due to other reasons, but the reasoning set forth in the AI Guidance is problematic on its face.
In contrast, steps (f) and (g) of claim 2 are found not to recite abstract ideas because they do not recite particular mathematical operations. Again, applying the AI Guidance, (f) and (g) could be found patent-ineligible by the addition of the mathematical operations involved in those steps.
The AI Guidance's “inside-out” approach will lead to counterintuitive patent-eligibility decisions, such as findings that claims narrowly reciting a mathematical operation are ineligible, whereas broader claims reciting the result of the mathematical operation are eligible.
Such determinations run afoul of the Supreme Court's admonition in Diehr that a claim drawn to statutory subject matter does not become nonstatutory simply because it uses a mathematical formula, and conflicts with the Court's exclusion only of claims that have far reaching preemptive effects.
Mental Processes
Example 48 identifies step (d) of “partitioning embedded vectors” in claim 2 as reciting a mental process, yet also concludes that the generation of a mixed speech signal in step (g) cannot practically be performed in a human mind. Thus, under the “inside-out” approach applied in the AI Guidance, claim 2 simultaneously recites mental processes and steps that cannot practically be performed in a human mind.
But instead of ending the § 101 analysis since the claim as a whole would not cover a mental process, the AI Guidance continues with the analysis. This result is contrary to the reasoning in Federal Circuit cases such as SRI International and in earlier USPTO guidance in Ex parte Hannun, which have found that claims that cannot be performed entirely in the human mind do not recite a mental process and cannot be an abstract idea.9
The USPTO Should Continue to Improve Its Guidance
The past 10 years of § 101 case law and examination guidelines have been anything but consistent, but there are some guiding principles that should prevail.
Exclusions to patent-eligibility under § 101 were always meant to be a “coarse … filter,” and the USPTO should adhere to this mandate in its examination guidelines.10 The current AI Guidance creates unnecessary confusion, but there are ways to implement the USPTO's newly issued guidance with the existing Supreme Court tests.
First, consistency is key: every § 101 analysis should be based upon a predictable benchmark, in this case, the identification of an abstract idea to which the claim as a whole is directed.
The new AI Guidance fosters inconsistency by instructing examiners to engage in a limitation-by-limitation approach first, which Alice and Diehr explicitly prohibit. A limitation-by-limitation analysis is instead more appropriate for the Step 2B inquiry of determining whether the claim limitations recite “significantly more” than the abstract idea.
Second, the USPTO should continue to solicit public comments for refining the patent eligibility analysis — especially regarding AI-related inventions — to ensure consistency with MPEP § 2103 (which states that “claims must be considered as a whole”) and with Supreme Court precedent. These public comments will signal to the Office that the recent guidance is confusing and will hopefully spur the agency to revise the guidance accordingly.
Third, the USPTO should increase its training for patent examiners on how to identify abstract ideas based on viewing claims as a whole. It is essential that patent examiners are well equipped to identify abstract ideas when appropriate, and to recognize claims that overcome § 101 rejections, to improve patent quality and uniformity.
Fourth, the USPTO could provide examiners in AI-heavy art units, for example, Art Units 2123 (Modeling and Simulation) and 2129 (AI), additional time to examine applications that may involve heightened § 101.
When faced with time constraints, it can be easier for examiners to maintain a weaker position on a complex issue than to articulate reasons that the claims are patent-eligible. Providing examiners with additional time to evaluate patent-eligibility issues carefully and methodically could pay dividends in improving patent quality of AI-related inventions as well as other inventions.
Ultimately, the USPTO must provide a predictable and objective test for both examiners and practitioners that is consistent with Supreme Court jurisprudence, to ensure that § 101 serves its purpose as a “coarse … filter” for patent eligibility and not a sieve.
Conclusion
The AI Guidance currently fosters uncertainty regarding patent eligibility under § 101, as it does not require identifying a claim's alleged abstract idea nor does it require evaluating the claim as a whole. Instead, the AI Guidance focuses on a limitation-by-limitation analysis that diverges from prior guidelines and from the array of § 101 cases such as Alice, Diehr, and BASCOM.
The USPTO should continue to refine its AI Guidance with procedures that adhere to Supreme Court precedent, and to more accurately evaluate AI-related inventions and improve patent quality. After all, examiners and courts should be aligned in their eligibility analyses to encourage innovation and ensure that AI-related inventions are not unnecessarily rejected.
Footnotes
1. Alice Corp. Pty. v. CLS Bank Int'l, 573 U.S. 208, 216 (2014) (citing Bilski v. Kappos, 561 U.S. 593, 611-12) (2010)).
2. See Rsch. Corp. Techs. v. Microsoft Corp., 627 F.3d 859, 868 (Fed. Cir. 2010) (opining that the section 101 patent-eligibility inquiry is only a threshold test) (internal citation omitted).
5. Mayo Collaborative Servs. v. Prometheus Laboratories Inc., 566 U.S. 66, 71 (2012).
6. Id. at 216 (internal citation omitted).
7. BASCOM Global Internet Services v. AT&T Mobility LLC, 827 F.3d 1341, 1344 (Fed. Cir. 2016).
8. Appeal No. 2018-003323 at 10 (P.T.A.B. Apr. 1, 2019).
9. SRI Int'l, Inc. v. Cisco Sys., Inc., 930 F.3d 1295, 1304 (Fed. Cir. 2019).
10. Rsch. Corp. Techs., 627 F.3d at 868.
Originally published by Westlaw Today
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.