Is the Banking Industry ready to face the AI challenge?
There's race between consumer companies trying to build financial services and financial services' providers trying to build consumer products.
Every year billions of dollars are spent on banking technologies and infrastructure initiatives each year. Althrough some banks have succeeded in implementing and scaling AI technologies, vast majority of them are still struggling to implement AI inside their banking systems. Many of the obbstacles hampering the banking indusry efforts, the most common is the lack of a clear strategy for AI. Their existing core technology kernel anf infrastructure becomes a challenge when it is weak are not updated.
Let's see some of the major challenges that the banking industry has been facing to implement AI and how that can be addressed.
An outmoded data backbone and operating structure
Some banks are adopting to start a smaller POC to test their legacy systems. Scaling these core/legacy systems becomes a huge challenge as the transaction queries build up and legacy systems can't handle them. The fear of the quality of data and accuracy of the results from the AI algorithms is high among the many technology leaders around the world.
This can now be reduced to a great extent by adopting cloud infrastructure. The cloud infrastructure enables scaling at a rapid phase and flexibility of platforms and services. The overhead of IT reduces and this enables the deployment of the smaller data marts which makes it easier for infrastructure management tasks and automation.
Lack of data
The biggest ongoing challenge for any AI based project implementation is data. In any banking system, data is usually present in different systems across different formats. Sometimes you end up having two redundant sets of data, both likely to be true. Combining or merging such customer's data into one single data warehouse for creating the system where your AI algorithm can learn from is now a challenge.
The best way to address this is to enable a best-in-class data management structure and API’s. This will make a 360-degree view across the various systems; internal & external and help make better analytic insights on the algorithms and models.
Talent/Domain specific knowledge/expertise
The banking Industry is similar to healthcare in this regard when it comes to the domain expertise. When the domain expert is not the right person, the end customer faces troubles that spoil the brand image and reputation.
The most advanced banking and finance industry still lacks an appropriate domain expert to handle the nuances of tax planning, investments, financial planning strategy and many more. Domain experts are the people who can make the AI project implementation a success by providing the reasoning and context to the data and algorithms.
Iterations and trust in AI
AI algorithms are built to learn over a period of time and not expected to work perfect from day one. This learning for the AI algorithm happens more effectively as you feed in more and more data, program a context for it. The AI algorithm then iterates itself with the data and starts learning for specific use cases. This is a long process, and this evolves with new information, as the system is fed with more domain specific expertise.
The AI systems and the programming interface can overcome these challenges by experimenting sandbox environments to test and refine applications to predict potential risks and subsequently decide on the infrastructure & technology to deploy at scale.
Risk, Compliance, Legal, Privacy
The complexity of data management across silos from multiple data marts are difficult to integrate with external applications and platforms. Furthermore, the cleansing and uploading (ETL) the data for diverse use cases such as BI, Analytics and machine learning has to be developed with appropriate controls and monitoring tools to ensure privacy, regulatory compliance & security. There are different rules and regulations across countries & states such as the local compliance laws such as GDPR in Europe, CASL & PIPDA in Canada, or CCPA in California.
So, the banking products and services launched in a new market has to adhere to the local regulations and that is not as easy as launching any other SAAS technology platforms.
Over the past decades, banks have been the first movers in adapting to new technology innovations and offering the best in-class technology solutions to their customers.
Mckinsey predicts the potential annual value of AI and analytics for global banking could reach as high as $1 trillion.
The AI-powered banking systems will enable better decision making for the end customers via capabilities that include advanced analytics and AI capabilities. Some of them include Natural language processing, Virtual agents, Facial recognition, Voice-script analytics, Block-chain, Computer vision, behavioral analytics and finally heading to a virtual or Augmented reality experience.