The concept of AI output insurance is also attractive to insurers in the professional indemnity sphere, as it allows for a shift from annual premium models to a per-question or per-matter model. This could lead to additional premiums adjusted in real-time based on potential error rate and liability. However, the author notes that this won't change the law regarding who can provide professional services. The article concludes by speculating that such insurance could become a core part of the offering of large foundational AI models.
Key takeaways:
- The article discusses the potential of 'AI Output Insurance' as a solution to the risks associated with the adoption and use of AI-based products by professionals, particularly in the legal and regulatory research field.
- The EU AI Act may enable this development by requiring extensive testing and data for AI products, especially those deemed 'high-risk' systems.
- 'AI Output Insurance' could shift the insurance model from 'premium per professional per year' to a 'premium per question asked or per matter dealt with' model, which could be more profitable for insurers.
- While AI is unlikely to replace professionals entirely, the availability of insurance against potential AI errors could reduce the need for human oversight and hasten the adoption of AI in professional practices.