EnglishJapanesePortugueseSpanish
Mohamed Lemine Beydia Discussed Key Use Cases of AI in Banking
Mohamed Lemine Beydia, a Data Science Manager at Deloitte attended Worldwide AI Webinar to discuss the key use cases of AI in banking. Read on for the highlights of his speech.
 
Check out his whole presentation on our website and YouTube channel.
 
 

4 barriers to AI adoption in banking

Mohamed started off by stating that AI is reshaping the world and banking is one of the industries that is being hugely influenced by it, though not every bank is in the same position.
 
There are four reasons why banks are struggling to adopt AI:
 

1. Lack of data

Mohamed cited a lack of data, siloed quality data, and outdated IT infrastructure as the biggest problem that banks need to solve first to embrace AI’s full potential. Moreover, there is a big problem when it comes to ownership of the data as data is fragmented between departments and it takes time to understand that there is no centralized way to access data.
 

2. Tighter regulations

Regarding regulations, stricter policies like the GDPR or data act have been preventing banks from using customers’ data however they want. 
 

3. Lack of talents

 Mohamed shared that a lot of young data scientists don't want to join banks at the beginning since there are problems linked to the data quality and the infrastructure. This challenge has made it harder for banks to attract talent. 
 

4. Lack of data literacy

According to Mohamed, there are a lot of bank leaders and executives who struggle with understanding the data and the power of data when data is well analyzed and well managed. 
 
 
 

Key current AI applications

Mohamed listed out some key current AI applications in banking such as Fraud and Money Laundering, Credit Decisioning, and Process Optimization to name a few. He then detailed a few key use cases that he believed banks should focus on.
 

Personalization

With the amount of data banks are getting from applications, websites, and the system itself, Mohamed claimed that they could leverage the power of AI for customer personalization. 
 
Using an AI model, banks can assess the visit frequency of potential customers, the products they were visiting on which device, and the location of said visitors and ultimately personalize their customer journey.
 
With AI, banks can directly show customers appropriate banners with CTAs, link customers directly with the call center, compare their behaviors to those in the past, and assign a probability of conversion.
 
Speaking from his own experience, Mohamed saw exponentially growing conversion rates growing exponentially when they had an accurate ML model that was able to categorize this kind of customer. The AI models could also do some intelligent targeting, meaning that based on the certain behavior this customer is having on a website, we might suggest a certain framework of communication.
 

Intelligent location

While intelligent location and geospatial data are not new in the banking industry, combining them with AI has shown a real improvement in different use cases.
 
Geomarketing is a prime example. With intelligent location, customer segmentation can be improved, test competitors can be monitored, customer proximity notification can be done and the result of a certain campaign can be measured accurately. 
 
Another interesting use case Mohamed witnessed was using the power of AI to identify a property attribute from satellite units to streamline the process of reducing the number of questions customers are asking since to acquire home insurance on a bank website, one’s house location must be given. 
 
With AI, banks can then directly get this code, locate the house, locate the neighborhood, use some open data, streamline these processes with the help of satellite images and give a more personalized experience for the customer and a more accurate estimation for the insurance.
 
Financial inclusion is another great use case. With the satellite image, banks can assess the performance and measure farmers’ credit risk.
 
Intelligent location can also be useful in assessing the real potential of your clients. Knowing where their customers live and having some open data allows banks to understand them better.
 

KYC & AML

With AI, we have seen some transaction monitoring where we use AI models such as graph analytics to discover non-obvious connections between individuals and transactions.
 
AI also helps with entity resolution where banks use some name-matching rules based on machine learning to represent their customers across national databases and international databases.
 
An ultimate beneficial owner is another important use case with KYC. Lastly, adverse media monitoring is where AI is engaged to screen the web social media to detect these people involved in manual monitoring and other criminal activities.
 

Personal finance management

From Mohamed’s observations, a lot of fintech companies have been building their entire business model on top of this. 
 
They would use an AI model to analyze their customer's transactions and behavior to give them some recommendations regarding their personal finance. With the help of speech recognition & NLP, anomaly detection, machine learning, and explainable AI, financial institutions can help customers limit expenses, put money in their savings accounts and ultimately retain customers.
 
Watch Mohamed’s whole keynote on our website and YouTube channel.