Intellectual property, artificial intelligence and contract law: are there any intersections and fruitful collaborations between these areas?

The article contains an analysis of the two main questions regarding AI and IP law: how shall input data used to train AI-systems and how should output results generated by these systems be protected by IP law? In the context of this analysis particular attention is paid on the role of contractual mechanisms used by private parties to resolve these issues in practice. It is concluded that in the situation of doctrinal and legislative conundrum contracts are often used by AI-operators (developers), AI-users, and rightsholders owning input training data to attribute IP rights to certain persons and to protect these rights. However, contractual mechanisms may not always be a good option. Regarding the issue of IP protection of input data contracts (in particular, licensing agreements) may be a helpful tool to get a permission to use data for AI training purposes and to address other related issues. Meanwhile, considering IP protection of AI generated outputs contracts usually serve as a means to monopolize these outputs by AI-operators and developers with no appropriate justification for this monopolization.