Cover Image

The rise of Machine Learning

April 19, 2021 - Reading time: 6 minutes

While the significant advancements in artificial intelligence in the first 2 decades of the 21st century have laid the groundwork for much of the future tech that is on the horizon and that we benefit from right now, A.I. ethics, and ethical A.I. choice-making in the context of systems that interact with human beings remains a lasting and perpetual concern. Ethics in artificial intelligence is in fact such a concern that many futurists often refer to a soon to be reached kind of technological singularity which is yet to occur — a hypothetical point in time in which humanity’s technological growth exceeds the confines of its original design and becomes an uncontrollable and irreversible entity of its own cognition, resulting in unforeseeable changes to human civilization.

As exciting and as rather startling as that may sound, the present day FinTech and RegTech industries of the modern economy are closely allied in being on the frontier of some of the trends that are driving this kind of technology forward, but they are also ensuring that ethics and humanity stay an intrinsic part of the A.I. equation.

Machine learning is a branch of A.I. premised on the concept that a system, by conducting its own data analysis, which then in turn automates analytical model building can learn from data inputs. It can then identify patterns in that data, and then use algorithms to make decisions with a minimum amount of human intervention.

It sounds quite spectacular, but as we all know, the real world has a great amount of nuance to it, which machines are not very good at accounting for. When machines make decision based on probabilities, they are not really experiencing reality but operating out of a set of parameters, if their environment changes and evolves to behave in an unexpected way the AI certainly can and will fail to adapt to this or comprehensively deal with such a change. And the complexity of A.I. means that it can be really hard to determine why or if a machine made a mistake, and that mistake can have serious consequences for human beings.

In an adapting and changing world, it is logical then to consider whether locking a system to remain affixed to a set of rigid parameters that cannot change is preferable, or otherwise alternatively rather to consider granting that A.I. the ability to evolve somewhat, and this indeed is what machine learning is and does. When it can evolve it effectively enables A.I. the ability to autonomously deal with changes in its operating environment and expand its parameters to absorb and internalise those kinds of operating changes. And this is where consequence management becomes a big issue.

The decisions made by machine learning systems have real world outcomes: it can influence investment profits and losses, express biases in risk preferences, influence hiring decisions that can have ramifications that vastly exceed just the make-up of who a company employs, determine who is granted a loan or gets access to a medicine and who does not, and even lead to automotive collisions simply due to circumstances beyond the understanding of an A.I. system. This type of technology raises an abundance of ethical and moral questions which are complex and not easy to answer. But one thing is for certain, and that is that this is the future, and it is driving change in our modern economies, and it is happening right now.

When RegTech companies for instance take a risk-based approach in executing their product functions, this is a very good example of how tech companies make use of machine learning to make these types of processes quicker and easier for consumers. By engaging this form of probability analysis and then reducing the risk on a vast scale, it can have positive benefits for consumers, making the application process for many kinds of services and products more accessible and even democratised. Given the concerns that exist in allowing A.I. to evolve and adapt, it is perhaps the RegTech industry that has found the most eloquent answer to the moral and ethical dilemmas that seem to exist in artificial intelligence systems.

RegTech companies do actually have a model to emulate, they have successfully automated, but they have still kept their humans. So when anomalies do occur and are expected, the human elements still remain to pick-up on what the A.I. systems cannot. It is fair to say in the broad conception of RegTech and machine learning, some adjudication is by all means still required to enable an efficient and practical business workflow to exist. And it is at this juncture that perhaps, RegTech has the answer to the broader A.I. problem. Keep the humanity where it is necessary, to make the decisions that should be made by a human.

Of course the complexities of A.I. systems are not getting any easier or smaller, 20 years from now the world will have changed at a dizzying pace, but if we are going to face a future of increasing automation and greater surveillance, and the worries of some kind of singularity occurring, then keeping the human decision making element will remain a vital piece of that equation.

#intergreatme #remotekyc #kyc # digitaltransformation #onboarding #knowyourcustomer #regtech