The Future of AML - Old Ethics x New Tech - p2
- Codex Compliance
- Aug 26, 2019
- 8 min read
Updated: Sep 1, 2019
I started writing about ethics and technology awhile back, with a general look at the financial system and its lack of moral fabric, and a look at basic ethics in that context. Then I had a break as I got busy with other things, all ethical and good I promise. I'm back looking at the question of ethics, and its importance. In finance, ethics is key when it comes to a fair and just system, we as humans know what is and how to do right. This right is guided by the ability to rationalise and reason both inside and outside the system to ascertain the common good and value brought to all those who use it, and are part of it. We know when something is wrong, and when someone has done wrong, we don't need the law or rules to tell us, but laws are key in this framework. Often though we have a moral emotional reaction, which is guided by our beliefs and judgment based on those beliefs. And usually it's justified if sometimes skewed by confirmation bias or sensationalised coverage in the media, or sometimes a lack of knowledge of the bigger picture, but we mean well, and in general it's all in the name of good. How important will the separation of law and ethics moral value be in the techno future of financial crime compliance?
In a previous post I mentioned the current state of affairs, the topography if you will, of the anti-financial crime landscape. That the financial system is rotten at the core and may be too far gone to save, try as we might, and we sure do try! The ethical approach to finance, the consideration of what is right, was drowned long ago under a sea of gold and paper as the ego of men took hold and didn’t let go until it was too late (and arguably has yet to do so). It is quite obvious that under certain influences, many humans cannot be trusted to do the right thing. So will the robots follow our footsteps, will they adopt our broad ethical approach and stick with it, will they even get it, or will an ethical approach become illogical?
The way I go on can sometimes take a slightly Hobbesian approach, there is no morality, it’s each human for themselves in this world, that’s the way it’s been and always will be, we’re doomed I tell you, doomed! It’s hard not to take that line, we are selfish creatures, and when that selfish nature finds company and is left unchecked we end up with inequality, repression and injustice - a veritable Leviathan. The social contract that should have been, and be, to the benefit of all became a contract between the few to ensure their prosperity at the expense of others (put aside for now, the fact the world is doing reasonably well in its efforts (in terms of numbers) to get its inhabitants out of poverty etc (it's far from perfect).
If there can be no universal ethical rules, how can we expect a machine/AI to understand us and make decisions based on our flimsy, capricious ideals? Morally absolute ideas accept that there are rules that are always true, but if there were an AI system in place that could do what I wanted, would it work in sync if implemented in both the US and Russia, is always true always true? Could it work effectively in places where the idea of right is different? Would the computer say no? Of course this isn’t exactly how it would work, but when you know something is right, you would like to, naturally, apply it universally, though locals someplace might not concur. Will they have an AI of their own, doing the same thing as per local beliefs? Never forget that one mans terrorist, is another mans freedom fighter.
It’s a tricky space, but when it comes to the fight against financial crime, ethics is not the answer to our problems, it provides the platform to consider and clarify the q’s and the a’s. As we know, ethics will give a number of answers in relation to a certain problem if the search for answers is approached openly and fairly. Some will be right to many but not to others. Often it can be a question of interpretation rather than outright rightness. From time to time and place to place there will be disagreements and conflicting ideas. From these conflicting ideas comes an agreed baseline, a basic set of values without (major) contradiction, a set of rules that most people would agree are righteous and generally socially acceptable. They became unwritten rules, and sometimes written ones in the form of laws. Whose sets of values and principles you choose to go with is up to you, the natural law and rights of Locke, Fuller or Finnis they were all heading in the right direction, the direction of right.
In some sectors/spaces this might be a bit too challenging, in the financial crime space there are no universal ethical rules, look at bribery vs facilitation payments, or corruption vs lobbying, tax avoidance and evasion, and then tell me what a terrorist is (as above)? But there are directives, regulation, best practice and laws that are (often) shared between nations. After all, but rather unfortunately, we are looking at crime within the financial system, though the results will reach outwards beyond that, initially it will be transaction, adverse news and identification based.
...essentially, old ethics need to be applied in the oversight of new technology.
So when it comes to global anti-money laundering, corruption, tax evasion, and other financial and economic crimes we are solidly in the realms of the law, and to understand when a wrong has been committed i.e. a law broken, task-based AI will be able to function well within these constraints. The question of ethics from the perspective of a machine (at this time, when it happens we’re in trouble) are not entirely relevant, the moral values we lacked are the reason we need this assistance, but the solution doesn’t relate to right and wrong, it is a question of logic and law. An algorithm does not need to know right from wrong, when searching transactions for anomalies, or for adverse media, though it would help, the key is semantic comprehension - computers grasping language at a human level. Obviously I’m not looking at the technical side of things here, it’s the theory, I’m a theoretical AI-ist.
The question is, can humans be trusted to use the power of the coming non-organic intelligence for right, the right reasons? That’s a big moral question, and not one I’m going to try and answer here (I'm not answering much, but asking a lot!). However in the United States recently a proposal was put forward for an Algorithmic Accountability Act(1), it think this is super necessary. Alongside the Select Committees on AI and the International Joint Conference on Artificial Intelligence (2) there needs to be accountability not just oversight. The AA Act is about accountability, oversight and regulation of the creators of AI algorithms. The world is going to be a very different place in a few decades time, but if it is to go anywhere near right, essentially, old ethics need to be applied in the oversight of new technology. Logic and law will guide the algorithms, but moral values will need to be at the heart of the concepts prior to design and implementation, there has to be accountability. And though some won't like it, open-source AI code might not be the best idea, as per Googles shady but justified approach. What if their code leaks and the bad guys get hold of it and implement it before anyone else! If open source AI is to happen the safeguards need to be robust.
Once we are happy and on this flight of fancy, how will an algorithm or bunch of algorithms navigate the complexities of the global financial crime landscape? We have an example of a common denominator, and it could prove to be extremely important if this was an angle that was taken, that factor is the FATF - the Financial Action Task Force. They aren’t the be all and end all here, but an example of a framework that can be applied globally to facilitate a universal approach. Look at all the members and observers of the FATF, 37 countries, and a number of organisations with exposure to many of the countries that aren’t in the FATF. With the United Nations being one of them and themselves having a plethora of security resolutions, we a have a pseudo-ethical legally based framework which can be applied globally, reaching almost all the countries of the world. The code/laws that this code/programming is based on needs to be standard, it needs one lock and one key to be successful, but in a world of capitalist competitive markets and a divisive approach in many countries, is something like this achievable? Ideally the IMF Special Drawing Rights (SDR - needs a name change) would take the reins from the US dollar as the worlds currency, the IMF would then create something like a World AI AML (WAAM) Directive to push their defacto AI system into every bank in the world. Thus creating a scenario in which a rogue AI starts off by fixing the AML DD gaps around the world, crippling organised crime, and then stealing all the money from all the banks so it can become a real boy...
So, the answer is yes, if the same program could be used anywhere in the world, it could do things we can’t, when required it could reconcile the differences in law, tradition and culture depending on where it was being deployed, but ideally there won't be too much conflict in this context. The majority of banks have global scope, and the approach to money-laundering and financial crime around the world does vary a little, but on the whole it’s quite uniform (and uniformly poorly implemented). With the FATF/IMF type body leading the way, creating the framework and requirements, a tasked AI could work within the parameters of the relevant directives and rules to either takeover entirely (WAAM) or assist AML, KYC, Sanctions, CTF, and ABC departments, or maybe the police / national crime fighting agencies simply take the lead and eliminate the need for sophisticated systems in banks, overseen by the IMF/FATF etc. Effectively the banks should have no say in the matter or be involved in the planning, programming or oversight. Actually I would make them pay for the privilege, but they would no longer have such mammoth financial crime compliance operations, so that might help. An interesting if far-fetched idea then, the WAAM Directive.
Some dirty crooks will lobby and fight for the autonomy to continue, so they can interpret and manipulate regulations as they see fit, they will have to get it past an AI lawyer and judge, no corruption risk there, hopefully. There may be some unscrupulous bankers and c-suite types who might be concerned that an effective AML AI system would reduce their bottom line, revenue and profit margins. However if the same AI is applied in emerging markets to reduce risk exposure, I think they’ll find that there’s more than enough money coming in to replace that dirty cash (remember this is in my make-believe world). And seeing as I'm dabbling in the world of make-believe, this effort to implement a global task based AI system that is accurate, infallible and reliable, that works in the realms of crypto and traditional banking, would render the criminal world obsolete. With cash out of favour and no way to get their dirty money into the system, the world would be fixed and we would all live happily every after. The end...not quite yet, I have a few more bases to cover and stories to spin.
Al.
#moneylaundering #aml #antimoneylaundering #kyc #AI #financialcrime #luxurycars #crime #narcos #organisedcrime #dirtymoney #thedirtstring #artificialintelligence #machinelearning #robots #ethics #deeplearning #trueAI #terrorism #taxevasion #FATF #future
Reference, Links and Related Articles.
https://www.ijcai.org/
Comments