5 Reasons Why it Won’t be Blockchain, But AI to Drive Financial Services
Utter the words ‘disruption’ and ‘financial services’ and your thoughts will be drawn to the bevy of technologies that were supposed to transform the sector. Artificial Intelligence (AI) and Blockchain are the most recent additions to the list, but this time around, they probably have the potential to drive real structural change. To explain their […]
Utter the words ‘disruption’ and ‘financial services’ and your thoughts will be drawn to the bevy of technologies that were supposed to transform the sector. Artificial Intelligence (AI) and Blockchain are the most recent additions to the list, but this time around, they probably have the potential to drive real structural change. To explain their potential and differences, Grant Thomas, Head of Practices at BJSS talks to Finance Monthly about the impact of these technology disruptions.
Blockchain, which was originally developed to support Bitcoin and other cryptocurrencies, is being heralded by the Financial Services industry as the next big thing because it supports peer-to-peer mass collaboration which could make many of the traditional organisational forms redundant.
In theory, Blockchain will reduce transaction costs – Santander expects to achieve savings of around $20 billion a year – so while the industry is still largely unclear on how it should be applied, there is a race to productionise it. Heavy Research and Development investments are being made.
The problem with Blockchain in the Financial Services industry is that it is largely pie in the sky. Its development landscape is being driven by a handful of large multinational organisations, mostly working as consortia, because they’re the only players able to handle its scale and apply the multi-jurisdictional experience the project needs. Open Source projects such as Openchain and Hyperledger are not sufficiently developed to offer a credible alternative. There is also a shortage of skilled talent available to build applications, or subject matter experts available to develop and validate business use cases.
AI on the other hand is far more mainstream. Companies such as Facebook, Google, Viv, and Nuance already provide frameworks and turnkey solutions, and AI technology is already being used by many Financial Services providers to handle everything from detecting fraud, to market regulation and customer interaction. The Royal Bank of Scotland, for example, has recently completed a trial of a ‘Luvo’ AI customer service representative to support internal customer-facing staff.
AI is capable of processing data to make decisions far more efficiently and accurately than humans can. It does this through self-learning to solve cognitive tasks. The technology crunches historical data and teaches itself to act based on the decisions that have been previously taken by humans. It also learns from its mistakes – so every time AI completes a transaction, it becomes more accurate.
The ‘disruption’ from AI comes from the efficiency savings that Financial Services providers will achieve by automating the highly-transactional jobs that are usually handled by humans. This will improve customer service quality and consistency and will improve both regulatory compliance and risk management. When they deploy AI tools such as IBM Watson, Financial Services organisations have both cost-cutting and customer satisfaction in mind.
The barrier to entry for AI is far lower than it is for Blockchain. There is ready access to experience, talent, and a burgeoning ecosystem to sustain innovation. That said, Financial Services providers should consider these five steps to ensure that their AI deployments succeed:
- It’s mostly about the data
Banks have large IT estates which generate a great deal of data – everything from customer demographics, to product adoption and market trends. There isn’t necessarily a requirement to collate data into a centralised data lake, but integration is important. Access to a self-service data model will allow, with minimum viable process, easy access to this data. Bear this in mind because providing as much data as possible is integral to the success of an AI deployment.
- Begin at the end
Look at the ideal scenario. Consider the outcomes that are to be achieved and reflect on the experience the user should have. Develop personas to keep users in mind, build models to ensure that business outcomes are being achieved. And only then, start to build the AI.
- It’s an elephant. Eat it slowly.
AI is huge. With an array of use cases as diverse as risk and fraud detection, customer relationship management, business development and cost reduction, AI is becoming increasingly important for financial services firms to remain competitive.
Don’t be tempted to tackle everything at the same time. When deciding which use case to start with, choose the lowest hanging fruit, build the AI, deploy and learn from it, and then finally, tweak it. Once this cycle is complete, move on to the next use case, applying the lessons learnt.
- Experiment in the Lab before moving to the outside world
Some organisations embrace the concept of Innovation Labs to generate new ideas for products and services. Others routinely use Labs as part of their project delivery objectives. Whichever way innovation is achieved, it is important to have a process, the right behaviours and lean thinking.
For AI, a lab provides a safe space for expose data, to apply simulations, to learn and to experiment with configuration tweaks.
- It’s not a project, it’s a journey
The project doesn’t end when the AI is commissioned – it continues.
A key part of disruption is the feedback loop. With the technology evolving quickly, this feedback mechanism should result in minor corrections being deployed quickly, while improvements are continuously implemented.