Erring on the side of caution with regards to companies collecting data isn’t a rarity, especially when it involves your finances. It’s worth noting however, that in order to create efficient, personalised and successful AIs requires the model to have as much data as possible in the system. Ensuring that the customer is aware of the uses of their data when it’s collected and how it’s protected will likely increase the trust each individual invests in the AI.
AI breakthroughs are ever present with DeepMind having taught its avatar to walk, and medical researchers honing in on the capabilities of medical imaging, and labelling and segmentation being near human perfect. It’s one thing to hear about all this groundbreaking research, but it’s another to apply it to a practical situation.
How are we utilising these progressions, and what does it look like when we train and adapt these models to function in a fully productionised system in a big corporate company?
This is exactly what Jakob Aungiers, head of the quantitative research development team at HSBC has been doing. He’s working with DL & AI for momentum strategies to help in a more tactical sense in areas such as anomaly detection and productionising the technology behind that.
In previous years, DeepMind have made astounding progress in AI research from teaching machines to win at ‘Go’, to their avatar learning to walk. Whilst a significant amount of their research has previously focused on games and such like, now that it’s being applied to real-world problems, Jakob explain that this could be huge. ‘Most of DeepMind’s ground breaking advances have come in the form of video games through reinforcement learning and teaching machines to generalise over all the different inputs and games that they are trained on.’ This ability to generalise has a huge application opportunity in FinTech and banking in general. The algorithms and models that are currently used in financial companies aren’t overly complicated as things are quite slow to move in the financial world and each model seems to be very bespoke to its set purpose, so to be able to employ models that can operate effectively on a variety of different tasks would be incredibly beneficial in finance. Currently there are distinct models for areas such as credit risk pricing, momentum and tactical signals, risk and investing, and so many more. A generalised model that could be applied over any kind of use case 'would open up a lot of opportunities to take the vast amounts of data that are available in this world to pump it through the model would enable it to produce the correct output without needing a team of quantitative researchers or data scientists.'
Jakob spoke about his work at HSBC in more detail in an interview at the Deep Learning in Finance Summit in London which is available to watch here.
As well as the use cases Jakob discusses that would benefit HSBC, banks and finance companies are benefiting from AI applications such as chatbots and AI assistants. These Ais are able to alleviate pressures of customer service whilst still providing a personalised experience for users. These bots are relatively inexpensive to develop (they can be built onto other platforms such as Facebook messenger so require less coding than standalone apps), and also help relieve the time consuming queries from being answered by valuable members of staff. Machine learning technologies are being used to predict the customer’s needs and provide relevant and personalised assistance. Bank of Scotland is currently using a virtual assistant to answer such queries and the iPhone app ‘ has been trained to handle customer questions about account balances and unknown transactions, how to make payments and what to do about lost or stolen cards.’
AI in finance can not only improve the efficiency of the corporation, but will improve the customer's’ experience turning tasks that were previously frustrating and time consuming into convenient transactions.
Keen to learn more about the impact of artificial intelligence in the financial sector?
Join us at the Deep Learning in Finance Summit in London next March 15 & 16 where we will explore the advances in deep learning tools and techniques from the world’s leading innovators across industry, research and the financial sector. Confirmed attendees include Swedbank, Royal Bank of Canada, Chevron, McKinsey & Company, CapitalOne and more.