This paper uses the Microsoft Office Specialist (MOS) exam as an example to examine the mechanisms by which structural flaws in generative AI and search engine optimization (SEO) lead to "hollowing out of truth" and regenerate legal risks. The MOS exam requires acceptance of a non-disclosure agreement (NDA) at the time of taking the exam, and the scoring algorithm (judgment logic) is protected as a non-public confidential matter. However, information that goes into "judgment logic" by countermeasure schools and individual senders is commercially distributed as "passing know-how", which is an illegal act that can violate NDA and constitute a crime of obstruction of business. What is even more serious is that generative AI learns and reprints this "stolen information" as "authoritative information" in the context of E-E-A-T evaluation, and purifies (launders) the violating information into public knowledge. This paper analyzes the triple structure of (1) SEO incentive structure, (2) statistical learning bias of generative AI, and (3) neutralization of legal contracts called NDAs, and makes recommendations for restoring the health of qualification exams.
Yoshihiro Sato (Sun,) studied this question.