Responding to an article about robotic inventors (Science, April 3, 2009), which reports about a robot that actually developed and tested hypotheses regarding certain yeast and discovered new gene sequences, we ask whether robotic “inventions” will ultimately become free to the public without the possibility for patent protection.

American patent law (35 U.S.C. Section 102) says that “[a] person shall be entitled to a patent unless . . .” (emphasis added), and proceeds to set forth a number of exceptions to patentability. But that preamble to section 102 limits the ability to patent to a person, probably not to a machine. This conclusion is reinforced by section 101 that limits the invention to the discoverer: “[w]hoever invents . . . may obtain a patent . . . .” Section 101 uses “whoever” not “whatever.”  Thus, a person using a robot that might make an invention may face some serious statutory impediments to patent protection.

The situation is compounded by Section 102(f) that states that one cannot obtain a patent if “he did not himself invent the subject matter sought to be patented.” Thus Section 102(f) prevents one from obtaining valid patent protection if he gets the idea in question — even in private — from another source.

There is some existing U.S. patent practice to offer some hope. Existing DNA/amino acid sequencing machines provide inventors with information that inventors later patent, of course. There is a difference because such machines are automated and not capable of cognition, and much of the inputs into such machines are provided and selected by humans. And the resulting data and results are analyzed and verified later by humans.

Another case in point we deal with frequently involves high-throughput compound screening to identify promising compounds for pharmaceutical, agricultural and other purposes, but again, the inputs into screening machines are human, and the outputs are analyzed by humans. Still another example we see involves Monte Carlo analysis using Spice for electronic circuit design, again involving human inputs. There are numerous similar examples. However, unlike our examples, the robots discussed in your article seem to have an independent ability to generate and verify hypotheses, perhaps leading in patent parlance to independent “invention” by the robot, not the human.

There is the possibility that the programmer of the robot could be the inventor if the robot were given the hypotheses to test and parameters to evaluate, in which case the human would probably be the inventor on the theory that the robot was simply the “hands of the inventor.”

We use Europe to illustrate how things might be different in other countries. Article 58 of the European Patent Convention sets forth the “[e]ntitlement to file a European patent application thusly: “[a] European patent application may be filed by any natural or legal person, or any body equivalent to a legal person by virtue of the law governing it.” That language — never applied to the robot situation in your article to our knowledge — seems to provide some wiggle room for the possibility of the robot being an inventor in Europe. But one still has to name the inventor on European patent applications, which leads to an interesting question. Would the robot’s central processor be listed as the inventor? If so, it might need to be identified by serial number and where it resides. Interesting possibilities.

If a robot were to be or become an “inventor” under the laws of the U.S. or Europe it would seem that the owner or lessee of the robot would probably be the owner of the “invention” rather as employers are generally the owners of employees’ inventions. However, owners or lessees of such robots should do something akin to what employers do with employees: still get solid written contracts from the developers of the robots to make sure robot inventions are owned by the owner or lessee.

One final thought: we might someday ask whether a robot that gains true “cognition” or self awareness should be considered a “person.”  That day seems a long way off and perhaps a bit too much for any court to decide now. But that day may be coming sooner than we expect.

Today’s post is by Guest Baristas Robert Stevenson, Esq., Joseph Murphy, PhD, Esq., and Thomas Clare, Technical Advisor, all of Philadelphia intellectual property law firm Caesar, Rivise, Bernstein, Cohen & Pokotilow, Ltd.

  Print This Post Print This Post  


  1. I think you need to look at Europe’s sanctions for getting the naming of the (human) inventor wrong, and who can put the matter in issue, to challenge the validity of the patent. In practice, one would name those who ran (whatever that means) the robot. Only the party who is the aggrieved rightful owner of the property in the patent can challenge errors in naming the inventor. In your hypothetical, who would that be? In other words, in Europe, robot inventors raise no interesting new issue.

    It continues to fascinate me, how the USA can simply ignore:

    1. The EPC emerged from intensive work of experts from England (common law) and Germany (with the same patent law as Japan and China).

    2. Thirty years of operation of the EPC has proven that its architecture works just fine, also in England.

    Why not just sign up to it, like the rest of the world already has? Ploughing your own furrow, on your own separate planet, hurts your industry more than anybody else’s.

  2. Robots that should be considered a person:
    when those person-robots become commonplace, ordinary, they will be PHOSITAs ; whatever they come up with will be obvious to them, so not patentable.

  3. […] Patent baristas there was a post the other that the caught my […]