The Basel Committee for Banking Supervision needs artificial intelligence, the human one does seemingly not suffice.

Sir it was with much interest, and hope, that I read Richard Waters’ report “Investor rush to artificial intelligence is the real deal” January 5. We sure need it urgently, at least in the Basel Committee for Banking Supervision.

First any reasonably good AI would most certainly not give in to emotions or sole intuitions as the Basel Committee did when for their risk-weighted capital requirements they decided that “risky” was risky and “safe” was safe. AI would see that in fact it is what is perceived as “safe” by bankers that which creates the biggest exposures and as a consequence the biggest dangers, if the ex ante perception turned out ex post to be wrong.

And AI would also be able to impose portfolio variant capital requirements instead of settling for Basel Committee’s “portfolio invariant” because as they admit when in “An Explanatory Note on the Basel II IRB (internal ratings-based) Risk Weight Functions” they explain: “Taking into account the actual portfolio composition when determining capital for each loan - as is done in more advanced credit portfolio models - would have been a too complex task for most banks and supervisors alike.”

And AI would also of course have asked about the purpose of the banks before regulating the banks… and therefore we would probably have saved us from the credit risk weightings that so distort the allocation of bank credit to the real economy.

That said we have to be careful though so that AI does not Frankenstein on us and imposes its own preferences (ideologies); like what the Basel Committee did when they decided that their bosses, the governments of the sovereigns, were infallible… and therefore banks did not need to hold any capital (equity) when lending to these.

PS. Perhaps we can have a competition between different AIs to see who comes up with the best proposal for how to regulate banks.