4 AI Use Cases in Fintech in 2018 according to PwC and Gartner
By Ishtiaque Hossain, Author of Wallowing in Dhaka. Ishtiaque Hossain is a writer from Dhaka, Bangladesh; currently working in the Malaysian FinTech industry.
Twitter Handle: @followishti
According to Gartner, by the end of 2020, 20% of citizens in developed nations will use AI for everyday operational tasks and a whopping 85% of CIOs (Chief Information Officers) will be piloting AI programs for their organizations through buy, build and outsourcing efforts. By the end of 2018, use of AI is predicted to be widespread through commonly used applications like cloud office suites.
A virtual model that can analyze and monitor physical or psychological systems, like predicting customer behavior (Digital Twin)
AI that is transparent and provides reasoning behind the recommendations made by it as opposed to “Black Box” AI that doesn’t provide reasoning (Explainable AI)
So where do all these AI trends fit into the Fintech industry?
There are 4 major areas where AI is going to make game-changing impact in the fintech industry.
1. Game-changing insights and explainable predictions
Firstly, the biggest use of AI in FinTech would be in generating insights that can accurately predict customer behavior. For example, there is already artificial intelligence in existence, that can learn your customers’ past behavior and make accurate recommendations on the customers’ credit-worthiness. Most machine learning algorithms existing today cannot provide reasoning behind how a decision was reached. PwC predicts that by 2018, Explainable AI will be widely adopted by enterprises either as best practice or requirement, and governments may make it a regulatory requirement.
2. Early detection and prevention of cyber-security threats
Security is a big issue, especially when it comes to online transactions. Using Generative Adversarial Networks (GANs), fintech companies can build robust security systems into their solutions. The GAN works with two opposing networks, one generator and one discriminator. The generator network creates fake data that looks exactly like the real data set. The discriminator network analyses both fake data and real data. Each network learns from the other and gets better over time. This system can be especially useful in detecting fraudulent behavior, suspicious transactions and early detection and prevention of cyber security threats.
3. Visual identification and verification
Using capsule neural networks to visually identify customers and documents could provide a huge leap in streamlining functions like account creation, loan and insurance origination and documentation. An AI could visually verify if the documents fed to it are authentic, and whether a customer trying to apply for a loan is the person who he or she claims to be. Coupled with a credit admin software that handles loan documentation for all parties (lawyers, evaluators, bankers), AI could provide degrees of automation that was previously thought impossible.
4. Chat-bots that are more “human”
Also, paired with a customer engagement platform, an AI could power next generation chat-bots that can intelligently answer customer queries, effectively reducing load from customer services department. Chat bots can be integrated with social networking sites, and accept requests for application and orders directly from social media channels. Gartner predicts that by 2018, more than 2 billion people will be regularly using conversational AI to interact with virtual customer assistants on smartphones and connected devices.