Deep Learning and Financial Regulation

Financial regulators are responsible for the public good. As the state is considered the “lender of last resort”, failure of a regulated financial institution may result in costs being paid by the public. As financial innovation accelerates the established cycle of crisis, bailout and reactively increasing regulation won't work. Autonomous regulatory agents (ARAs) will change the regulatory paradigm and break this cycle. Driven by deep learning, the complex regulatory web will be reduced to first principles reflecting what constitutes the common good. A regulatory framework mediated by ARAs, with decisions driven by deep learning, will not create contradictions in the first instance. Over time, ARAs trained by financial regulators will gain significant insights. ARAs will also be used by financial institutions to ensure regulatory compliance. ARAs must be capable of fully explaining their decisions.