There is a type of project trying to address the core pain point of AI systems—untraceable decision-making. Their approach is to introduce zero-knowledge proofs, allowing autonomous agents not only to make judgments but also to provide mathematical proofs to endorse the entire process. It sounds a bit complex, but essentially it is about opening the "black box" of AI, so that each decision can be verified, ensuring transparency and credibility. This solution has potential applications in fields like financial risk control, data processing, and more, especially when you need to prove the reasonableness of decisions to regulatory authorities or partners.