AI offers the capability to make intelligent decisions or accurate predictions through algorithms within a specifically given and defined problem domain.
There are different types of AI — Narrow intelligence, General intelligence & Superintelligence
- Narrow AI is catered towards specific tasks — such as new recommendation on Facebook and Google, classifying spam emails on Gmail, product recommendations on Amazon or Netflix, or Home value forecasting on Zillow, chatbots, flights or hotel price forecasting on Hopper or Google Flights.
Top 4 Artificial Intelligence stocks to buy and why | Data Driven Investor
Artificial Intelligence ( AI) continues to make its way into the mass consciousness in 2019. And through applications…
- General AI covers a broad spectrum of problems across different domains and industries. It would be capable of understanding the world as we humans do have the same capacity to learn how to perform a wide range of tasks — it’d involve social skills, creativity, wisdom, science, engineering, literature.
In theory, an artificial general intelligence could carry out any task a human could, and likely many that a human couldn’t. This is the ideal state.
- Super AI can surpass human intelligence and everything that humans can do as well as everything that computers can currently do. It can think about abstractions which are impossible for humans to think.
Currently, AI builds its solutions using the following approaches:
Why are banks using AI in their back-office functions? Is this a logical transition?
Banks deal with a lot of data day in and day out — client data, demographic data such as name, email address, home address, age, income, etc. and then financial data — investments, transactions, mortgages, credit cards, etc.
To better serve their customers as one integrated bank, banks want to aggregate this data in one place and derive insights out of it to create contextualized hyper-personalized strategies for their clients.
Using AI reduces the time spent in analyzing client data, finding patterns out of it and AI makes it possible to generate near-accurate insights about a particular cohort of clients or a client itself.
So reducing time and cost is definitely one success metric in using AI in banks. But also, how can banks offer the maximum value to their individual clients, rather than generic advice/recommendations on products.
There are other applications of just automating processes and forecasting/predicting etc often include:
- Predicting the price of investments/stocks
- Fraud detection in credit card activity
- Forecasting the home value in a particular neighborhood
- Sending you the best or cheaper offers according to your spending patterns
How is AI being used in payments/transaction banking?
- Credit Scoring — Assessing an individual’s creditworthiness is a critical function of any bank or financial organization. ML algorithms are useful to convert customer data into a credit score, which banks or credit unions can use. NLP and AI can improve both that process and its outcome.
- Fighting fraud — Through machine learning algorithms, payments companies can analyze more data in new and innovative ways to identify fraudulent activity. Every consumer transaction includes uniquely identifiable information, and with AI and machine learning, payments companies can search rapidly and efficiently through this data beyond the standard set of factors like time, velocity and amount.
- Personalization — Organizations can leverage customer data from different sources to provide specific contextualized insights to their customers. This also makes it possible to tweak their solutions in response to the customer, advise and provide recommendations to make certain life decisions based on the specific customer’s situation.
- Data intelligence / Sentiment Analysis — Financial institutions and organizations can make better business decisions by collecting insights from the data around them — news articles, blogs, forums, social media, financial influencers as well a host of customer data. Identifying entities and patterns from this data, finding clusters and correlations between the different types of entities and hence finding relevant deeper contextualized insights can help businesses identify topics of discussion in the market, get ideas for trading, or discover events that might impact their investments.
- Document search and analysis — Natural language processing can be used to automatically read and understand documents that involve loan or mortgage processing. The AI can then comb through several thousands of these documents to extract and summarize the most relevant information from them.
- Chatbots — With the ability to respond to questions through Natural Language Processing, by understanding the direction and objective of the question, it is being able to provide an accurate or a near-accurate answer in natural language back to the customer.
- Computer vision- powered Retail and Payment transactions — AKA what Amazon has done with it’s Amazon GO retail stores. There are recognition sensors throughout the store that can detect what people are picking up and charge them for the items without having to go through the traditional checkout process.
What impact does AI have on efficiency, earnings and customer experience?
- AI is increasing efficiency across the entire value chain — right from the time involved in getting payments processed, identifying and preventing any fraudulent activity, and then providing personalized insights to the customers based on their transaction history.
- This automation, in turn, reduces the cost to the organization, because less time and resources are spent in performing routine tasks, and the manpower is better utilized performing specialized activities to further enhance the customer experience.
- Customer experience is made more valuable — customers/clients are able to navigate the platforms more easily, are able to get quick support and answers to their questions and are able to get personalized help and offers from the company. Which is an immense win from the traditional ways of serving the customer.
What does the use of AI in banking mean for the customer?
A lot. A few thoughts where AI in banking would benefit customers include:
- Receive hyper-personalized and contextual help/advice/suggestions as well as recommendations/offers from the bank instead of generic ones — this would provide immense value to them as users of the platform or the service
- Checkout seamlessly in any retail store without the need to wait in a queue for cashiers to check them out — this would provide a definite time efficiency to the customers
- Be protected from any fraudulent activity on their accounts and hence would be able to better trust their financial institutions
How does the experience change for the customer?
- The customer’s interaction with the product or the service does not change, however now after using AI you might find efficiencies here and there. You as a customer might feel the product or platform is speaking to you, that things are getting done a bit faster, my queries are getting addressed sooner, I’m able to provide feedback to the platform and it’s actually improving
Do customers need to learn any new skills?
- The beauty of AI is that it happens in the background. The customer experience enhances but whichever channel you choose to access the product or service (mobile or web or through a retail front), AI runs in the background without you even knowing about it.
- So no particular skills that customers need to learn, but they should definitely practice patience in the initial stages — eg. whilst interacting with a chatbot that is newly launched, it’s still learning from the customers and the data, hence some amount of patience and trust is needed in order to allow AI to flourish.
Do you need to give customers reasons for using AI?
Yes and no. Consent is needed. But beyond that, there are other considerations, such as to provide explanation around what is being done with the data, how is it getting used, what part of the data is being used. And then the output of the AI algorithms — eg. insights or recommendations or answers should be explainable i.e. companies should be able to explain how those answers arrived at, etc.
How is the use of AI being regulated? How are banks helping shape this?
With the explicit rise and awareness of GDPR and PSD2, banks are making it a point to be privacy compliant.
- Explicit consent mechanisms need to be created and maintained for any apps/platforms that use big data and AI
- Explainable AI — banks and fintechs are making it necessary to explain the results spewed out of an AI algorithm
- Creating audit trails — Banks are trying to maintain an audit trail surrounding the use of AI and its decisioning that can be thoroughly explained to regulators. This audit trail must be monitored to make sure that AI is producing understandable outcomes and isn’t being used where there isn’t sufficient reason or experience to rely on it.
Advances in AI technology in the past few years and expected in the future
- Adversarial attacks or Generative Adversarial Network (GAN) — train two neural networks on the same dataset and make them play a so-called “real or fake” game. Dueling neural networks open an opportunity for data scientists to create entirely synthetic datasets that can be used for training machine learning models. Different neural networks challenged each other to learn to create and improve new content in a recursive process.
- Combination of supervised and unsupervised based learning
- Deep Reinforcement Learning
Will AI replace the existing workforce?
Companies will always need more human resources. However, they must also invest in technology to help the existing workforce be as efficient as possible so they can pursue the work with the highest impact. In addition, AI investments will help organizations serve our audiences leading to higher efficiencies and reduced costs.
However, it’ll be critical to put in some kind of human oversight on AI. Personalization for the clients is critical, but it does not mean just handing over financial and advice judgements to algorithms …