Suppose an original piece of software contains an AI algorithm (that is NOT patented) that tries to optimise the solutions the software produces.

Out of the box the software does not infringe any patents, but after learning from user input and goals it starts homing in on a behaviour which has been patented.

It 'learns' that the patented solution is a optimum solution.

Would that be an infringement?


It doesn't matter. The US patent system is based on "first to invent", so it doesn't matter if other inventors find new ways to discover an identical solution later. It is still infringing on the original patent.

If you think about it, computers (and AI) are no more or less a tool of invention than a pencil and draft board. There is nothing in intellectual properly law which absolves real people from infringing on prior inventions based on "my robot came up with it".

  • Thanks, that's more or less what I imagined. But I don't understand who would be liable for the infingement; the orginal software supplier supplied the AI tools, but with no knowledge of the eventual solutions it would find, if they were liable then it means anybody who supplies AI could become liable for violations of things that may not even have been patented at the time of software release; if the end user is liable for infingements....well that's a mess! – Roger Irwin Dec 20 '18 at 8:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.