By Robin Hammond
From artificial leaves to digital rodent brains, computer scientists are successfully implementing processes from the natural world. Does the Alice ruling on software patentability provide a glimpse into how these technologies will be treated in the future? In 15 years, will computer-implemented natural processes be patent-ineligible?
The utilization of natural processes in computing is astounding. Recently, researchers in Melbourne developed a solar-power device that is close to simulating photosynthesis. The Energy and Environmental Science journal published these details in an article; in the article, researchers claim “This simple and adaptable system addresses key criteria for the large-scale deployment of an artificial photosynthesis device” to “provide a cheap, clean and renewable source of hydrogen as the ultimately sustainable fuel.” In August, news broke that IBM had created a digital equivalent of a small rodent’s brain, a type of chip with neuro-synaptic architecture, called TrueNorth. Both the chips and subsequent software recreate a brain’s neurons and synapses. The TrueNorth chip is another leap forward for computing through ‘deep learning’. Deep learning operates in two stages: “First, companies … must train a neural network to perform a particular task. If they want to automatically identify cat photos, for instance, they must feed the neural net lots and lots of cat photos. Then, once the model is trained, another neural network must actually execute the task. You provide a photo and the system tells you whether it includes a cat. The TrueNorth, as it exists today, aims to facilitate that second stage.”
It has taken many years, but the courts are finally addressing the realities of software patentability. In Alice, the Supreme Court did an about-face in terms of the validity of business-method and software patents. No longer are ideas or processes patentable because a computer executes them. While the opinion did not itself mention software, it has heralded in a new age of hostility at the U.S. Patent and Trademark Office toward business-method and software patents, with many courts overturning patents under the new standard of Alice.
In 15 years, will natural processes that are merely implemented by computers be novel enough to qualify for patents? Or will the two-step inquiry applied in Alice be similarly interpreted once these processes become commonplace? The analysis under Alice begins with the question: does the invention consist in significant part of a patent-ineligible concept—for example, a law of nature, natural phenomenon or abstract idea? If so, the invention is patent-eligible only if the remaining parts of the invention have an “inventive concept”—one or more elements that ensure a patent on the invention amounts in practice “to significantly more than a patent upon the ineligible concept itself. In Alice, the patents at issue were for computerized escrow. The court decided that the abstract idea of escrow was patent-ineligible and performing escrow on a computer was not sufficient to provide an inventive concept: “A mere instruction to implemen[t an abstract idea] on … a computer … cannot impart patent eligibility.”
Does the shift in the patentability of software patents in Alice foretell the future of the patentability of artificial nature? One could foresee a future where the idea of brain functioning (arguably a natural phenomenon) is in itself not patentable, and the mere implementation of specific neuro-synaptic architecture by a computer would not rise to the level of inventive concept. Or will future courts embrace Judge Rader’s broad interpretation of the applicability of §101? Where natural laws are restricted to “universal constants created, if at all, only by God, Vishnu, or Allah”; “even gravity is not a natural law“; and the slightest bit of labor mixed with nature is property under Lockean reasoning.
Image source: http://maxmania.pl/nauka/syntetyczny-lisc-wytwarzajacy-tlen/.