I previously explored how AlphaEvolve can become the future to automated innovation across every thinkable industry where you can setup a strong and fast feedback loop. Now consider the following:
Imagine you have an automated innovation system within X. X is a defined field with a great deal of prior art in the form of patents. Now your automated innovation system is powered by an objective function that optimizes for a problem relevant metric, and for patentability or patent whitespace. In addition, you bolt onto that a system for generating the actual patent applications for the discoveries that pass your threshold criteria.
This would be patent trolls on steroids, allowing for complete blockage of innovations within given fields, completely saturating a given field’s IP space.
This could allow companies to employ strategic and pre-emptive blocks of IP space. The field need not even be profitable yet, only strategically valuable in the future. This could create moats that are almost impossible to navigate around.
Again it becomes clear that we are dealing with some technological shifts that moves much faster than our human processes.
Rethinking incentives
Having the patent offices applying AI systems to deal with the flood of applications such an automated innovation system could produce would be counter-productive, and lead to an arms race and downward spiral. Maybe it is time to rethink the entire incentive structure behind patents.
The traditional bargain of patents: temporary monopoly in exchange for disclosure no longer makes sense once systematic innovation can generate thousands of new ideas that would take humans decades to explore
I don’t know what the solution is, but I do know that this is a problem worth thinking about.