Thinking Machines & Patentable Inventions

(Getty Images)

‘A patentable invention is a mental result’ (Smith v. Nicholson, 88 U.S. 112)

The purpose of this note is to examine whether there is an existing possibility of generating innovative and patentable mental results by a software intelligent (SI) entity. In order to recognise SI capable of creating mental outcomes this note sets conditions for a modified version of a Turing Test[1] to study SI's ability to generate patentable invention equivalent to, or indistinguishable from that created by a human being. Aiming to contribute to the modelling of a legal definition of a SI actor as an entity capable of mental outcomes resulting in a patentable invention and as such deserving legal recognition. It is appropriate to issue certain caveats here; this note does not attempt to argue that SI is able to think.

The US Supreme Court held that a patentable invention is a new and industrially applicable mental result[2] that is valid only when represented in some physical form.[3] Furthermore, the inventive act must consist of two stages: conceiving of the idea and reducing of that idea in practice. If ‘the mind is no more than the aggregated functions of the brain’[4] then we may argue, that a machine is capable of producing mental states by simple virtue of the fact that such a machine already exists in the form of the human brain. This note asserts further that the mind is the software of the brain,[5] which becomes an interactive ‘computational machine’[6] that receives precepts from outside environment, processes that information, and takes actions accordingly. In line with the above, it may be argued that the Court via not distinguishing between human or artificial mental results left open to further interpretation the possibility of patentable invention being generated by any type of mind, including artificial/software one.

This casts doubt, however, on whether there is an artificial brain that is capable of producing both stages in generating patentable invention, which are: (a) mental results and (b) their physical representation. It might be suggested that if only software mind is capable of generating its mental results it might have the necessary quality to represent its mental outcomes in a physical form. With this in mind, the question is not about whether the machine is capable of thinking the question is rather whether the machine is able to create mental results and its physical representation that are identical to, or indistinguishable from, that generated by a human being.

This takes us back to the question, whether a software intelligent actor could create mental outcomes in terms of whether the machine could pass a Turing Test.[7] As indicated by this test, if a justice was confronted with two different mental results and their physical representations, one belonging to a human being and the other one to software intelligent actor, without being able to distinguish between them, then the SI entity would pass the test. Suggesting that those outcomes of an artificial mind satisfy the substantive conditions of patentability and make SI an actor capable of producing patentable inventions.

References:

(Oscola type of referencing)

[1] Alan Turing in his paper ‘Computing Machinery and Intelligence’ (1950) 236(1) Mind 433, 433-460; argues that because 'thinking' is difficult to define, what matters is whether a computer could imitate a real human being. Alan Turing created a trial to test artificial entity’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human.

[2] Smith v Nicholson 88 US 112.

[3] Clark Thread Comopany v Willimantic Linen Company 140 US 481.

[4] VG Ivancevic and TT Ivancevic, Computational Mind: A Complex Dynamics Perspective (Springer Science & Business Media 2007) 92.

[5] N Block, ‘The Mind as the Software of the Brain’ in DN Osherson and EE Smith and LR Gleitman (eds) An Invitation to Cognitive Science: Thinking (MIT Press 1995) 377.

[6] AD Redish, The Mind Within the Brain: How We Make Decisions and How Those Decisions Go Wrong (Oxford University Press 2013) 147.

[7] Turning Test (n 1)