Friday, February 10, 2017
The Constitution tells us that patents can be given to “inventors,” and the Patent Act states that protection is available to “[w]hoever invents or discovers” an invention. These are not generally controversial propositions, but like so many legal regimes, technology is forcing these analog laws to deal with digital phenomena. The culprits here are artificial intelligence and software capable of inventing new technologies. Can patents be given to digital “inventors,” and if not, does any human have the right to patent such an invention?
Obvious comparisons can be drawn to whether non-humans can be “authors”—as required by the Constitution—for copyright purposes. For instance, can a digital composer of music be given copyright protection for its work? The academic consensus is that technology is not an author (for Constitutional purposes), but the agreement dissolves from there. Some have argued that programmers should be given ownership rights—a reasonable proposition—but this sentiment is far from universal.
With little guidance from copyright law, parties have looked elsewhere for ideas in the patent sphere. It has been posited that—if a non-human cannot be an inventor—current patent laws require the first person to “discover” the value of the non-human invention to be the inventor. This may be correct, but one must consider if that policy creates maximum incentives to further the progress of technology. Perhaps an amendment to the Patent Act is appropriate.
There are several potential avenues to address this issue. Ownership could be allocated to the programmer, the company owning the hardware, or to no one at all. Granting ownership to any party incentivizes creation of more inventing software and artificial intelligence. This benefit, however, comes at the price of granting 20-year patent monopolies at a relatively small marginal cost to the patentee (after software is accounted for). Does this encouragement to create and use of inventing software/computers come at too high of a cost to society?
Denying any patent protection enriches the public domain by including all non-human inventions—a social positive. That policy, however, potentially discourages the creation of inventing machines (as it decreases their market value) and incentives fraud before the Patent Office. Should a party identify a valuable invention created by a non-human, it might falsely claim that a human was the inventor to secure a patent. This would benefit dishonest firms at the expense of honest companies and society at large.
Lastly, there is an interesting potential distinction to be drawn between inventing algorithms or software (e.g., genetic algorithms) versus artificial intelligence. Non-human corporations continue to obtain greater rights under the Constitution (see Citizens United). Might artificial intelligence (e.g., something that could pass the Turing Test?) be recognized as an inventor at some point, while a “mere” algorithm or software might not? The question is probably premature, and as presented, is likely not sufficiently nuanced. However, the topic may eventually be raised.
At this time, no one has the answers to the above legal and policy questions, though I’m sure many commentators (myself included) will chime in. I invite readers to voice their opinion in the comments.