Since current intellectual property laws are ill-equipped to protect neural network architectures and training algorithms, there’s a growing number of companies increasingly relying on trade secrets. But is this for the best? Many people argue that this undermines transparency and contradicts the patent system’s fundamental purpose of encouraging technological disclosure, and the lack of transparency itself could invite further regulations. There is of course valid logic behind this behavior. The extensive disclosure required in a patent application poses a significant risk, especially when the innovation, such as an algorithm, is relatively easy to replicate, and the pace of technological advancement in the domain far outstrips the slow-moving bureaucracy of patent offices. These challenges raise the question of whether the AI revolution calls for an IP revolution – to protect innovation from being stifled, but also to maintain relevancy of the patent system. What could a new IP regime where disclosure is actually promoted look like?

In this episode, which was recorded in our new fancy RallyCast studio at our London office, Carl and JP discuss this topic and more, such as:

• Why should IP be considered a growth catalyst rather than just an asset protector?
• Why is IP such an underutilized source of intelligence? IP data can inform decisions across a wide range of areas, from investments and R&D strategies to product development and even partnership building. How can we make it more accessible to different stakeholders?
• Do we need to speed up the processing of AI protections, and what would the potential outcomes be? By the time a patent is published after 18 months, most AI companies will have advanced far beyond what was filed.
• Is the EU AI Act something that addresses a symptom rather than the root cause?

We also get a book recommendation from Carl, namely Deep Utopia by Nick Bostrom. Carl explains why it's an impressive piece of work.

Learn more about IPRally at iprally.com and get a free, 3 day trial of IPRally at iprally.com/trial.