A Science letter by Robert Stevenson focused my attention on the eventual patentability of new "automated" discoveries. This is of course a letter in response to the "automation of science" previously reviewed. Apparently, it should be legally difficult to patent any invention made by a robot: the American patent law strictly refer to the inventor as "a person", while the European law seems more broad. Thus, supposing a brilliant robot scientist is ever build, no man might protect/deserve those inventions for its proprietary benefit. Legacy is not considered: A invents robot/algorythm B, B invents drug C, but C is not patentable. Honestly, once developed such a robot, it should take no longer to develop a second robot mimicking human creativity (i.e., writing good and bad dates and results on a paper lab-book). Are we so close to Singularity? Interestingly, in a previous Science paper, Debra Meloso from the italian Bocconi University has modeled the patent system and proposed a better way to promote intellectual discovery (maybe including generation of robot-scientists) that should be based on a sort of 2.0 trading of discoveries. Might a machine sell a product? Ask to lawyer Crawford.
Stevenson, R., Murphy, J., & Clare, T. (2009). Robot Inventors: Patently Impossible? Science, 324 (5930), 1014-1014 DOI: 10.1126/science.324_1014a
Meloso, D., Copic, J., & Bossaerts, P. (2009). Promoting Intellectual Discovery: Patents Versus Markets Science, 323 (5919), 1335-1339 DOI: 10.1126/science.1158624