CyborgIntell’s Post

Now that we are meeting with clients, partners and potential clients in so many countries across the globe (+-30 at last count) a growing topic of conversation around AI is risk management, regulation and potential regulation, and resulting compliance challenges. This trend is particularly true with banks, insurers and lenders. We come across organizations who have operationalized +- 1000 AI/ML models that they know of (we suspect, based on experience, there probably many more - especially if those running at vendors and partners are included). How can these organizations be sure that every model is performing optimally, is transparent, fair, ethical and without bias. In addition are these models compliant with various regulations, are there version controls and audit trails and therefore clear accountability. If all of these are in place, can they be easily proved to risk committees, regulators and auditors? It is for all of these reasons, and also so that organizations can trust extensive use of AI, that CyborgIntell has built our Model Risk Management capability on our AI/ML platform. We believe that real-time explainability at both a model and a decision level, transparency, auto-documentation, version control, inventory, tracking, reporting and audit trails are essential. In addition there is comprehensive functionality to save time, money and reduce risk. Some of this includes automatic real-time measurement of scores of KPAs, with notifications of model or data drift, root cause analysis in minutes, champion challenger capability to test performance of existing models. Use of AI/ML is only going to intensify - we recommend that organizations implement best practices around risk and compliance as early as possible.