All Those in Favour Say AI

By Carlos Somohano, Big Data Evangelist, WHISHWORKS

AI is not a new concept in the banking industry. Even as far back as the 1970s, trading floors have been using automated trading systems, which with the help of complex algorithms, were able to make millions of trades per day – what is known as ‘high-frequency trading’. Still, it took almost three decades to have the computing power and storage capacity to evolve the technology in a commercially viable way.

Fast-forward to today, a recent MarketsandMarkets report states that the market size of AI in Fintech is expected to grow from USD 1.3 billion in 2017 to USD 7.3 billion by 2022

This alone shows that increasingly, financial institutions are understanding the potential of AI technologies and the value it can bring not only to the organisation but also ourselves as their customers.

However, we don’t expect to see any radical change in the near future. At the moment, AI is being used to enhance existing back-end processes and non-AI automations that will provide significant efficiency gains for the operation of the banks with financial performance, risk and costs being the main drivers. In the front-end, we will experience these changes only incrementally, primarily in the form of improved customer experience and transaction security.

Times are changing  

Fraud prevention is one of the main areas that banks are investing in. Machine Learning (ML) and Deep Learning (DL), both offsprings of Artificial Intelligence, are being used to improve security by using algorithms to compare vast amounts of data from many different sources and assess the likelihood of a transaction being fraudulent. The difference with the non-AI solutions currently used, is that ML and DL programmes learn and adjust their algorithms according to past outcomes as well as the changing habits of the account owner, achieving higher levels of accuracy.

Graph Analytics is another very important development achieved thanks to AI and increasingly used for fraud prevention. With Graph Analytics banks can analyse data networks and identify activity or relations that may signify the existence of organised crime groups. 

AI is also gaining momentum within risk and compliance. With the ever-tightening regulatory environment and directives like MiFID II and Dodd Frank that aim to increase market transparency and investor protection, banks are facing significant fines and reputational damages if they fail to provide evidence. In 2017 alone, these fines amounted to £230 million. AI-based solutions are being used to search and compare both historical and real time data to identify abnormal behaviours and transactions. This way AI solutions act as a prevention mechanism rather than their non-AI counterparts that search only through historical data to produce results after the fact.

Another area where we see increased application of AI solutions by banks is in customer service. Chat bots and conversational interfaces are increasingly used to facilitate the communication process and address simple customer enquiries, reducing customer waiting time to be served, whilst allowing more time for human advisors to deal with complex problems. Even when not used in the front-end of customer service, AI solutions can help financial advisors to make better, more personalised recommendations to customers by weighing previous account activities against current data, significantly reducing the risk of mis-selling.

There are many other tactical processes, where banks are introducing AI-based solutions to optimise processes and reduce risk. Some of the projects we have seen being implemented include Emotion Analytics for Claims Management, Prescriptive Analytics within Management Information Systems, Satellite Analytics for Construction and Property Finance, Document Summarisation for Legal Departments and many more.

Positive outcomes 

Financial fraud last year exceeded £1 billion in the UK, with online banking fraud increasing by 226% and telephone banking fraud by 178% year-on-year. Security and fraud prevention are the areas where we expect to see the biggest positive effects not just for the banks but most importantly the general public. In addition to the anomaly-detection AI-based applications currently used to detect fraud, we start seeing added security measures on the user side, including facial recognition, voice recognition, or other biometric data.

Fighting cyber-crime

A controversial area for AI has been cyber-crime. Many cyber criminals already use AI thereby having an ongoing advantage as their bots become more sophisticated with every instance making them one of the highest priority threats globally. The cost of cyber-attacks to UK businesses last year was over £40 billion and banks represent a prime target. The cost of cyber-attacks to banks however is much more than the reported figures if one considers the cost of having to shut down operations until the attack is addressed and then get the services back up and running, the cost of upgrading their security measures hurriedly and, perhaps the biggest of all, the reputational cost. To protect their assets against cyber-attacks, banks are now deploying AI to monitor and detect anomalies within their systems and networks and block any abnormal behaviour before it causes any damage. Taking it a step further, today we are also seeing the uptake of Generative adversarial networks (GANs), a class of AI algorithms that use unsupervised machine learning to train two different networks one against the other, an attacking one simulating cyber-attacks and a defending one to protect the bank. The two networks try to outsmart each other and in doing so continuously becoming smarter and more efficient. This way the defending network will be more effective in battling real cyber-attacks.

The robots are coming 

As with most technologies, AI is not a revolution; it has been evolving over many decades and it continues to do so as we speak. Today, AI is creating new job opportunities, as new skills are required to develop and manage different AI-based applications. In parallel, as the adoption of AI has been happening gradually, other positions are up-skilling or morphing to include AI components within their particular domain. In the end however, it is inevitable for the need for certain skills to start diminishing and for new roles to emerge.

Having said that, with organisations trying to optimise their operation to remain competitive and profitable, it is important for every one of us to make sure that we too refresh our skills in line with the prevalent trends and technologies within our area of expertise, so we too remain competitive within our organisation and the job market in general.

By Carlos Somohano, Big Data Evangelist, WHISHWORKS

AI is not a new concept in the banking industry. Even as far back as the 1970s, trading floors have been using automated trading systems, which with the help of complex algorithms, were able to make millions of trades per day – what is known as ‘high-frequency trading’. Still, it took almost three decades to have the computing power and storage capacity to evolve the technology in a commercially viable way.

Fast-forward to today, a recent MarketsandMarkets report states that the market size of AI in Fintech is expected to grow from USD 1.3 billion in 2017 to USD 7.3 billion by 2022

This alone shows that increasingly, financial institutions are understanding the potential of AI technologies and the value it can bring not only to the organisation but also ourselves as their customers.

However, we don’t expect to see any radical change in the near future. At the moment, AI is being used to enhance existing back-end processes and non-AI automations that will provide significant efficiency gains for the operation of the banks with financial performance, risk and costs being the main drivers. In the front-end, we will experience these changes only incrementally, primarily in the form of improved customer experience and transaction security.

Times are changing  

Fraud prevention is one of the main areas that banks are investing in. Machine Learning (ML) and Deep Learning (DL), both offsprings of Artificial Intelligence, are being used to improve security by using algorithms to compare vast amounts of data from many different sources and assess the likelihood of a transaction being fraudulent. The difference with the non-AI solutions currently used, is that ML and DL programmes learn and adjust their algorithms according to past outcomes as well as the changing habits of the account owner, achieving higher levels of accuracy.

Graph Analytics is another very important development achieved thanks to AI and increasingly used for fraud prevention. With Graph Analytics banks can analyse data networks and identify activity or relations that may signify the existence of organised crime groups. 

AI is also gaining momentum within risk and compliance. With the ever-tightening regulatory environment and directives like MiFID II and Dodd Frank that aim to increase market transparency and investor protection, banks are facing significant fines and reputational damages if they fail to provide evidence. In 2017 alone, these fines amounted to £230 million. AI-based solutions are being used to search and compare both historical and real time data to identify abnormal behaviours and transactions. This way AI solutions act as a prevention mechanism rather than their non-AI counterparts that search only through historical data to produce results after the fact.

Another area where we see increased application of AI solutions by banks is in customer service. Chat bots and conversational interfaces are increasingly used to facilitate the communication process and address simple customer enquiries, reducing customer waiting time to be served, whilst allowing more time for human advisors to deal with complex problems. Even when not used in the front-end of customer service, AI solutions can help financial advisors to make better, more personalised recommendations to customers by weighing previous account activities against current data, significantly reducing the risk of mis-selling.

There are many other tactical processes, where banks are introducing AI-based solutions to optimise processes and reduce risk. Some of the projects we have seen being implemented include Emotion Analytics for Claims Management, Prescriptive Analytics within Management Information Systems, Satellite Analytics for Construction and Property Finance, Document Summarisation for Legal Departments and many more.

Positive outcomes 

Financial fraud last year exceeded £1 billion in the UK, with online banking fraud increasing by 226% and telephone banking fraud by 178% year-on-year. Security and fraud prevention are the areas where we expect to see the biggest positive effects not just for the banks but most importantly the general public. In addition to the anomaly-detection AI-based applications currently used to detect fraud, we start seeing added security measures on the user side, including facial recognition, voice recognition, or other biometric data.

Fighting cyber-crime

A controversial area for AI has been cyber-crime. Many cyber criminals already use AI thereby having an ongoing advantage as their bots become more sophisticated with every instance making them one of the highest priority threats globally. The cost of cyber-attacks to UK businesses last year was over £40 billion and banks represent a prime target. The cost of cyber-attacks to banks however is much more than the reported figures if one considers the cost of having to shut down operations until the attack is addressed and then get the services back up and running, the cost of upgrading their security measures hurriedly and, perhaps the biggest of all, the reputational cost. To protect their assets against cyber-attacks, banks are now deploying AI to monitor and detect anomalies within their systems and networks and block any abnormal behaviour before it causes any damage. Taking it a step further, today we are also seeing the uptake of Generative adversarial networks (GANs), a class of AI algorithms that use unsupervised machine learning to train two different networks one against the other, an attacking one simulating cyber-attacks and a defending one to protect the bank. The two networks try to outsmart each other and in doing so continuously becoming smarter and more efficient. This way the defending network will be more effective in battling real cyber-attacks.

The robots are coming 

As with most technologies, AI is not a revolution; it has been evolving over many decades and it continues to do so as we speak. Today, AI is creating new job opportunities, as new skills are required to develop and manage different AI-based applications. In parallel, as the adoption of AI has been happening gradually, other positions are up-skilling or morphing to include AI components within their particular domain. In the end however, it is inevitable for the need for certain skills to start diminishing and for new roles to emerge.

Having said that, with organisations trying to optimise their operation to remain competitive and profitable, it is important for every one of us to make sure that we too refresh our skills in line with the prevalent trends and technologies within our area of expertise, so we too remain competitive within our organisation and the job market in general.