Guardrails removed spam, off-topic, unclear, or duplicate replies.
Ask a question below.
Published answers will appear here.
Totally agree that “independent AI audits” can provide a false sense of security--and that’s exactly the problem IVOs are intended to address. In our original 2019 paper proposing the regulatory markets approach on which the IVO model is built, we discussed the credit rating agencies explicitly as a failure mode and cautionary tale. Critically, the credit rating agencies are NOT effectively regulated by the government; they are insulated from liability for their ratings. The expert on this (whom we cited) is law professor and securities regulation expert @FrankPartnoy. Here’s what we said in 2019: A core lesson is clear: regulatory markets cannot operate effectively if they are not effectively overseen by government agencies with budget and the capacity to resist capture. Without strong oversight, private regulators may only ‘compete’ by lowering their standards. (Evidence of this is provided by the example of S&P, which recovered from a negative shock to its reputation–caused by errors requiring withdrawals of ratings–by issuing more optimistic ratings than its competitors (Baghai & Becker, 2019).) Our proposal rests on the design of effective outcomes-based regulation of private regulators. The IVO model is trying to get us as close to the regulatory markets model as feasible now, given political and economic realities. The regulatory markets model is a fundamental innovation in regulatory methods, and we need to build capacity and expertise we haven’t built before. We’re all on the same page if we’re focused on making sure “independent auditors” are themselves accountable to public oversight. https://x.com/ReadTransformer/status/2075640916740559155
If you’re really interested in what we should learn from the credit rating agencies, read the expert here: https://buff.ly/zSTIl4b
Totally agree that “independent AI audits” can provide a false sense of security--and that’s exactly the problem IVOs are intended to address. In our original 2019 paper proposing the regulatory markets approach on which the IVO model is built, we discussed the credit rating agencies explicitly as a failure mode and cautionary tale. Critically, the credit rating agencies are NOT effectively regulated by the government; they are insulated from liability for their ratings. The expert on this (whom we cited) is law professor and securities regulation expert @FrankPartnoy. Here’s what we said in 2019: A core lesson is clear: regulatory markets cannot operate effectively if they are not effectively overseen by government agencies with budget and the capacity to resist capture. Without strong oversight, private regulators may only ‘compete’ by lowering their standards. (Evidence of this is provided by the example of S&P, which recovered from a negative shock to its reputation–caused by errors requiring withdrawals of ratings–by issuing more optimistic ratings than its competitors (Baghai & Becker, 2019).) Our proposal rests on the design of effective outcomes-based regulation of private regulators. The IVO model is trying to get us as close to the regulatory markets model as feasible now, given political and economic realities. The regulatory markets model is a fundamental innovation in regulatory methods, and we need to build capacity and expertise we haven’t built before. We’re all on the same page if we’re focused on making sure “independent auditors” are themselves accountable to public oversight. https://x.com/ReadTransformer/status/2075640916740559155
Guardrails removed spam, off-topic, unclear, or duplicate replies.
Ask a question below.
Published answers will appear here.