Customers in the financial sector have particularly high expectations and requirements when it comes to solutions based on Artificial Intelligence, which is revolutionizing the banking industry with intelligent chatbot-based solutions. Machine learning is a powerful weapon in the fight against threats and risks in the banking sector. We spoke to Piotr Kubica, a machine learning specialist at JCommerce, who works for ING Bank Śląski, about the challenges and opportunities offered by Artificial Intelligence algorithms.

– In the United Kingdom, two-thirds of banks use machine learning solutions, and solutions based on Artificial Intelligence plot new courses for the development of the banking sector around the world. According to Garner’s report, machine learning is one of the main technological trends which are worth paying close attention to in 2020. Like any other technology, machine learning brings enormous opportunities, but also challenges. How does machine learning differ from traditional programming, what threats can it help to eliminate, and how does this technology build new standards in banking? We discussed this with machine learning developer Piotr Kubica, who is developing a solution to be used by ING Bank Śląski.

Let’s start with what probably interests financial sector representatives the most. What in your view are the main benefits of using machine learning in banking?

– I think that surely the basic goals that all companies strive for are maximizing profits and reducing costs. Machine learning can help in both cases. The most popular use of machine learning at the moment is personalization in advertising: the goal is to maximize profits by increasing sales, suggesting products tailored to specific user needs. In addition, machine learning helps to minimize costs by automating repeatable processes.

In the context of automation – we often come across the opinion that someday everything will be automated… Will it be possible to automate all processes in the future?

– I think that until we have this general Artificial Intelligence that will be able to do everything (in the way that humanoid robots are depicted in all Science Fiction films) – machine learning will simply be based on optimization, mathematics, figures and on how these programs and algorithms are written. At the moment, people still determine the quality – the technology is only as good as the people who have created it. However, this area is constantly evolving and over time we will be able to solve more complex problems. But will we ever be able to solve all of them? I don’t think so.

What are the expectations of the banking and financial sector in terms of machine learning? You have mentioned how important cost optimization is.

– I think that the banking sector expects a lot from machine learning. Currently, all organizations strive to be “data-driven”. For the banking and financial sector you have mentioned, the regulations are also a significant aspect. There is the Polish Financial Supervision Authority, which right now does not say yes to everything. Machine learning is so new that sometimes the algorithms used are not easily understandable.

In which situations will machine learning work?

–For example, machine learning is great for decisions on granting loans. However, the regulator must first agree to use machine learning. Currently, traditional credit risk modeling, traditional assessment, statistics and rule models are used for this purpose. However, machine learning is also used in many other areas of banking and finance – for example in the personalization of offers, which we discussed earlier on. It is very important to provide customers with a suitable offer.

So it is mainly the marketing aspect?

– Not only that. I think that currently it is the most developed part, but right now solutions based on text processing, e.g. document handling automation, are also developing strongly. I think that the automation of customer service processes through the use of chatbots and voicebots will also be the future.

How does Artificial Intelligence support banking consultants? Here I mainly mean customer service.

– The bank uses a chatbot solution that allows customers to solve problems by themselves without having to contact a human consultant. It is possible to get answers to some simple questions that often come up. There is no need to read the regulations or the Frequently Asked Questions section. I think it can also work the other way around. Consultants can use this kind of knowledge base when the customer calls the helpline. They do not need to have particularly detailed knowledge as they can find information in a faster way during the conversation with the client, and thus the service becomes more effective.

And when it comes to customers’ User Experience?

– The bot cannot answer all the questions, certainly not non-standard ones. Some people, however, do not realize that they are talking to a bot, because the chatbots’ answers are prepared in such a way and are so good that it’s impossible to tell the difference by asking simple questions.

This is interesting in the context of Turing tests. Many companies have struggled to pass them successfully.

– A year or two ago, the Google AI conference took place. There was quite a popular film on the Internet, in which the bot was making an appointment at the hairdresser’s for a client, a smartphone user. We could see the Google assistant calling and arranging such a visit, knowing the specific plans, and having an insight into the calendar. It turned out great, everyone at the conference applauded. However, now information is emerging that 70 percent of calls are reportedly being tracked or taken over by people anyway. So it’s all controlled or taught during the testing phase. It’s not as perfect as everyone would like.

How does ML support security processes and all processes aimed at preventing data theft or identifying users?

– There are many possibilities when it comes to using machine learning for user identification. Each user has their own unique behavior on the site. It’s a behavioral approach that allows others to identify the user. I do not necessarily see the potential in data protection alone, but I do see the potential in user identification – creating a model for each user and checking how a person behaves in different environments using a telephone or computer. Machine learning would be able to distinguish user data. The question is, how well would such a thing work?

The behavioral biometrics you are talking about is an interesting issue. Biometric banking is also often used. The customer calls the bank to unlock an account and the system can recognize their voice. How does it work?

– This is an interesting issue. We each have a unique fingerprint, iris or voice. Machine learning is also used to recognize them, to the best of my knowledge. In the event that we want to unlock our account using voice recognition, we must first teach the system the model of our voice, so that later it will be able to ascertain with a high degree of probability that we are the person we claim to be.

We have talked about the benefits and possibilities of machine learning. What are the biggest requirements and threats in this area?

– There are also some threats, as machine learning also has some weaknesses that can be exploited. But here the war on reinforcement will begin again – or even just begin. Just as there are IT safeguards, so too will safeguards against using model gaps exist.

What model gaps do you mean?

– What can prevent you from training a model based on the voice of the customer, which will be able to impersonate him at the bank and recreate his voice based on a few minutes of conversation? IBM has already published research stating that they are able to replicate someone’s voice based on a mere five minutes of voice material. In fact, something similar probably took place in September 2019 – I mean social hacking by means of Artificial Intelligence, where millions were extorted.

How has it come to this?

– Artificial Intelligence learned and copied the voice of a company’s CEO and persuaded the vice-president of another company to transfer some money to another account. This is a kind of arms race, a fight for better and better models. So there is a model that learns to recognize your voice, and then we create a network that learns your voice and impersonates your voice. Subsequently, a model is created that will later distinguish your artificial voice from your real voice.

Machine learning and data collection also raises questions about whether our data is secure.

– It depends. Machine learning itself uses data that must be stored somewhere. ML models do not store data, but store certain ways, rules, weights or divisions of decision making. Also, the moment of data storage is important here – this is not my area of expertise, you would have to ask data storage security specialists. In the learning process, however, we use available data that is secured, and later all application development processes already have to comply with these security requirements.

What are the real threats and how do we counteract them?

– There are interesting examples of how you can cheat on the neural network by changing one pixel in the image. There are many possibilities, many things that you need to pay attention to while modeling. But surely a lot of things will arise as this field develops. Because, as you mentioned, this industry is quite new, so there is a lot to discover – not everything is precisely defined, e.g. how machine learning processes should be created. There is no single, rigid pattern – and that’s why it’s interesting. As part of the work of a machine learning developer, it is very often necessary to sit down and think about how to do a given thing well, how to prevent the leakage of relevant information.

When it comes to the use of machine learning in banking – what would you personally consider to be the greatest achievement in this area?

– Theft prevention. This issue often comes up. Some users have negative experiences with this though. They try to pay for something by card and it turns out that their card is blocked. Then there is a phone call from the bank: “Is that really you making a transaction?”. And this is an example of an unsuccessful use of the system.

Bad User Experience. However, in good faith.

– It all happens in good faith to prevent someone from using your card for a nefarious purpose. Previously, rule models were used here. It is about such cases when someone is shopping in Katowice at about 4pm and in Moscow around 6pm. It is impossible to cover such a distance and get around so quickly by plane.

And now these methods are no longer used?

– It seems to me that hybrid models are coming in – which also represent a highly interesting solution. The combination of machine learning with rule models, where machine learning allows us to draw upon some additional cases that people could not define or lacked some vital knowledge.

What does machine learning offer in such situations?

– In these cases, machine learning can detect additional scenarios that were not taken into account. But it can also limit just the kind of cases that we discussed earlier on – meaning that these rule models can have too many so-called false positives. That is, cases in which the model thought that a theft had occurred, when in practice it had not. It also means a reduction in the number of cases that are incomprehensible to a user who is trying to make a transaction. The purpose of blocking a bank card is to prevent theft – which is certainly not pleasant. By way of comparison, it is certainly “less pleasant” than unlocking the card later.

Thank you for the interview.

Interlocutor:
Piotr Kubica

Machine Learning Engineer at JCommerce. He is keen on solving problems, thereby creating added value for the client. He co-creates innovative machine learning-based solutions for the companies from the FinTech sector. In his spare time, he likes to taste good coffee and is keen on playing computer games (mostly RTS). An amateur of windsurfing and skiing.

Dodaj komentarz

Comments:

CONTACT US!

Would you like to learn more about the possibilities of cooperation? Do you have a question? Write to us!

I hereby agree that JCommerce Sp. z o.o. shall process my personal data (hereinafter ‘personal data’), such as my name, surname, e-mail address, telephone number and company name, for commercial purposes.
I hereby agree that JCommerce Sp. z o.o. shall process my personal data (hereinafter ‘personal data’), such as my name, surname, e-mail address, telephone number and company name, for marketing purposes.
I hereby agree that JCommerce Sp. z o.o. shall process my personal data (hereinafter ‘personal data’), such as my name, surname, e-mail address, telephone number and company name, for recruitment purposes.
I hereby agree that JCommerce Sp. z o.o. shall process my personal data (hereinafter ‘personal data’), such as my name, surname, e-mail address, telephone number and company name, for future recruitment purposes.
I have been informed by JCommerce Sp. z o.o., 3 Ks. Piotra Sciegiennego St. 40-114 Katowice – the personal data controller – that: - The provision of the aforementioned personal data is voluntary but essential for commercial purposes if I have chosen a request for proposal, or recruitment purposes, if I have chosen the remaining options;
- I have the right to access the content of my data, including to receive copies of it and correct it, delete it and limit the processing of it, as well as the right to transfer it, the right to object to the processing of it, and the right to withdraw my consent at any time. However, the withdrawal of my consent shall not affect the lawfulness of processing carried out on the basis of the consent in question prior to its withdrawal;
- A statement of withdrawal of my consent to the processing of personal data should be submitted to the headquarters of JCommerce Sp. z o.o. or sent to the following e-mail address: zgody@jcommerce.pl. The withdrawal of consent to the processing of personal data shall result in the inability to fulfil the aforementioned processing purposes;
- The personal data provided shall be shared by JCommerce Sp. z o.o. with the company’s authorised employees and individuals collaborating with JC under civil-law contracts, who are involved in the implementation of the purpose of the processing;
- The data provided shall be processed on the basis of the relevant provisions of Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation), ‘GDPR’;
- Should you have any questions regarding the protection of your personal data, please contact us by e-mail: odo@jcommerce.pl;
- The personal data provided shall be processed for the purpose for which it was supplied, or until I express my objection in this regard. In the event of filing an objection, JCommerce Sp. z o.o. shall no longer process my personal data for the aforementioned purposes, unless it can demonstrate that there are valid and legally justified grounds overriding my interests, rights and freedoms or my data is necessary to establish, pursue or defend a claim, if any;
- I have the right to file a complaint to the supervisory authority if I consider that the processing of the aforementioned personal data violates the provisions of the General Data Protection Regulation of 27 April 2016.