This is a topic that often comes up across social media and conferences that focus on FinTech developments within retail financial services industry. There is no doubt with the fast growing digital ‘fourth revolution’, we are witnessing rapid build and deployment of tech throughout industry products and services.
A good place to start could be to define what AI is. The ‘father’ of computer science Alan Turing published a paper in 1950 entitled Computing Machinery and Intelligence which can be viewed as the first time AI was considered as Turing asked the question “can machines think?” From this point Turing offered a test (known as the Turing test) where a human interrogator would try to distinguish between human and computer text response.
We then had Artificial Intelligence: A modern Approach written by Stuart Russell and Peter Norvig, which is viewed as a leading AI textbook covering four AI goals across; Systems that act and think like humans and the ideal approach for systems that act and think rationally.
AI simply put is a field which combines computer science with robust datasets to enable problem solving. There are two AI strategies to consider:
- Rules based: Produces pre-defined (deterministic) outcomes that are based on a set of rules coded by humans. This system is called as simple AI which uses two components; a set of rules and a set of facts.
- Machine learning: Defines its own sets of rules that are based on data outputs. This system is based on a probabilistic approach (generate a catalogue of all possible events) thus machine learning can provide a comprehensive picture for risk management.
Both the above can bring great benefits to industry as we wrote in our blog; The luddite fallacy and the rise of AI in retail financial services.
Differences between rules-based AI and machine learning
- Machine learning models require far more data than rules-based systems. Rules based systems can operate with a simple data driven framework.
- Rules based can be design for specific tasks that need an algorithm to aid decision making e.g. specific compliance and risk management tasks. Machine learning constantly evolve, develop and adapt in accordance with training using statistical rules, thus over time can offer more in depth and granular management information
- Rules based systems are immutable (unchanging) whereas machine learning systems are mutable. Rules based will require human intervention to change the algorithms, with training, machine learning can transform over time.
As the FCA reporting in their paper Machine learning in UK financial services, we are witnessing an increased use for AI whether it be rules based or machine learning. Governance, risk and compliance management, algorithmic trading, robo-advice, marketing, client relationship management, secure transactions, process automation are just a few.
As another ‘father’ of science (behavioural economics) discussed in his latest book ‘Noise’ algorithms are an important part for overcoming the high and hidden cost of inconsistent decision making. Therefore, dependent on what service or product you are offering, each type of AI discussed can play an important role in delivering accurate data and save time and costs. So AI can play an important role within financial services and support (not necessary replace) human to human engagement.
Please click the below icon link to MO®'s platform and learn more about MO® today..