AI Trends and Applications for Finance and Technology in 2021

Artificial intelligence is revolutionizing the way financial institutions and tech companies use their data to automate repetitive tasks and gather valuable insights. Some popular examples of AI trends and applications in Fintech are fraud detection, risk assessment, or virtual financial assistants.

At the Data Science Salon for Finance & Technology December 8-10, we had the opportunity to chat with leading Fintech data scientists and ask them about their favorite use cases of AI, the power of AI in – 19 challenges for COVID AI 2021 trends, and challenges to successfully implement AI projects.

Spoiler Alert: Contains advice for data scientists working in finance and technology. Let’s Dive!

Which use case of AI in finance and technology are you most excited about?

Money laundering exposure is a key challenge for financial services firms, and the risks are regulated and trusted. Money laundering is the illegal distribution of money in a legal system and the dumping of large sums into undetectable or fake accounts. More sophisticated algorithms are now needed to manage these risks as criminals can easily exploit loophole-based guards and develop their own money laundering techniques. To keep up with changes in the regulatory environment and criminal typologies, compliance teams will need to apply new technologies to improve their detection and reduce analyst analysis. This is an interesting use case where AI can improve existing compliance processes by developing different card-based algorithms, not only in compact sub-charts but also in the fact that money laundering involves cash flow through banking services in the chain of accounts. ; It is a major challenge to make the detection of money laundering very accurate. –Debasmita Das, Senior AI Specialist at Mastercard

“While I want to spend some time on the problem, I’m still interested in using AI to detect and fight financial fraud. The most direct impact on the institution, as quick responses and accurate identification, were found real, leading to immediate savings on fraudulent damages, but also provides systems that respond quickly and intelligently to customer behavior to provide timely feedback on fraudulent activity. Create respect for customers. Implementation of AI maintains quick decisions, while each customer allows for personal decisions. “

‘There are actually so many use cases that are very interesting when it comes to FinTech. Especially now that blockchain technology is knocking on our door, many things happen automatically and predictions are being made at an exponential rate. Even with augmented reality in the photo, use cases like Google Glass are very interesting. –Vijay Pravin, data analysis specialist at Siemens Germany

How has AI helped Fintech meet the challenges of COVID-19?

“Organizations grappling with the problems of COVID-19 can benefit from automating more manual labor so they can do more with less. AI models that automatically respond to support cards, remove spam from inboxes, or identify valuable customers to focus on can save valuable time and increase its impact. –Kristie Wirth, computer scientist at Zapier

“COVID-19 has created tensions in many industries, and while many financial institutions were able to adapt, many people were not that happy. Any event that puts people under pressure, especially financial stress, emphasizes financial institutions. Intelligence is based on customer service, with implementing automated chat services, assisting the service team, and identifying and alerting users in the first place, it also provides answers to customers and faster solutions to their problems.

“With COVID-19 in place, most traditionally operating industries are enjoying tremendous success. The move to digital will not only help the organization but also the customers and clients who will attend virtually the majority of meetings or events.” –Vijay Pravin, a specialist in data analysis at Siemens Germany

What is the biggest challenge you have encountered in applying AI in finance and technology?

The financial industry faces an ongoing challenge to apply AI to a number of regulatory steps being taken regarding the use of customer models and data. The challenge, however, is not to find ways to deal with regulations but to find ways to manage our responsibility to use data and models in an ethical way. Regulatory measures are a type of arrangement with some degree of accountability, and while they may be larger for financial institutions, they are not exclusive to funding. Ultimately, anyone using AI must understand the responsibility to use it ethically and support benevolent ideals for technological development. –Jeff Sharpe, Senior Manager / Technology Leader at CapitalOne

‘Money is something that people care a lot about:

That’s why security breach prevention and ease of use have always been a challenge in this particular industry. And data protection laws (e.g. GDPR – General Data Protection Regulations in Europe) that apply in different countries are also a point of analysis when it comes to delivering real-time AI applications. –Vijay Pravin, a specialist in data analysis at Siemens Germany

One of the biggest challenges is the explainability of neural networks:

to avoid legal complications, every financial institution must be able to support its decision based on a rational AI model for auditors, regulators, and stakeholders. While there are approaches such as SHAP, which are designed to decipher the logic behind complex neural network decisions, many financial institutions prefer tree-based algorithms or linear functions because they are easy to interpret. The second challenge is to address bias in artificial intelligence algorithms due to historical bias. Dealing with prejudice and justice based on ethical and administrative considerations has become essential to the ideal implementation of AI. The third biggest challenge is to expand the proof of concept (POC) before the introduction of the AI   model. The reduction in the size of the POCs creates a series of barriers such as sanctity and data quality; support of interested parties; sufficient funding and fear of failure if the model does not deliver the desired results. –Debasmita Das, Senior AI Specialist at Mastercard

Translate »