The fastest industry I've seen evolve so far is that of generative AI. Capabilities are skyrocketing week on week, and everyone is (or should be) wondering what the implications are for the work that they do. Generative AI refers to a class of machine learning algorithms that can create new data based on existing data. The most buzz has been around art generation and code completion, but this technology can also enable the development of highly personalized and tailored financial solutions.

Banks are losing their differentiation as each year passes. It's no longer about features but benefits, everything that used to be a feature - mobile banking, real time notifications, card controls - is now table stakes for every fintech. With banking being viewed as a utility, consumers will start expecting increased personalisation in line with social networks and other experiences. Generative AI offers the potential to analyze vast amounts of customer data, predict financial trends, and generate bespoke solutions for individual clients.

Understanding Generative AI

Before we dive into how generative AI will change finance, let's understand what it is in the first place. Generative AI focuses on creating new data instances, content, or solutions based on existing data sets. These data sets can be very broad or very niche, it depends on the use case. The most popular tools we've seen lately, like ChatGPT and Midjourney, have been of the broad type. Advanced machine learning and deep learning is used to learn from the data, make predictions and generate new information without being explicit told how or what to do.

Machine learning learns from data by looking at patterns and trends and by creating models that can be used for predication or classification tasks. Deep learning, on the other hand, is a different beast. It's a much more advanced form of machine learning that using articifical neural networks to try and mimic what the human brain does: take in a lot of information and make decisions and predications based on it.

To get bit more technical, there are several key components of generative AI:

  1. Generative Adversarial Networks (GANs): GANs consist of two neural networks – a generator and a discriminator – that are trained simultaneously. The generator creates new data instances, while the discriminator evaluates the quality of these instances, determining whether they are real or generated. This adversarial process helps both networks improve over time, resulting in the generator producing increasingly realistic data instances.
  2. Reinforcement learning: This is a type of machine learning where an agent learns to make decisions by interacting with its environment and receiving feedback in the form of rewards or penalties. Reinforcement learning enables the agent to learn from its actions and improve its decision-making capabilities over time, ultimately leading to the generation of optimal solutions.
  3. Variational Autoencoders (VAEs): VAEs are a type of deep learning model used for generating new data instances by learning a probabilistic mapping between input data and a lower-dimensional latent space. They are particularly useful for generating complex data structures, such as images or text.

Now that we've covered a bit of what generative AI is how it works, let's take a look at how it could matter in the financial industry.

Applications of Generative AI in Fintech

Analysing customer data

"Data is the new oil" has been a saying for over a decade now, but it feels like we haven't been able to extract a lot of value from that data. If I go to Amazon and buy a pan and a set of kitchen knives, I don't want to be shown another pan or more knives - I want to be shown pots or oven gloves or cookbooks.

The better artificial intelligence gets, the more value we'll start extracting from our data. Using generative AI we should be able to identify more patterns and trends hidden within the data and get a deeper understanding of customer behaviour, preferences and needs. This all leads to better, more personalised financial services.

Customer segmentation: better categorisation of customers based on data that includes various factors like demographics, financial behaviour and preferences. This leads to more relevant marketing campaigns, tailored financial products and happier customers.

Fraud detection: with a deeper understanding and processing of the data, unusual patterns can be spotted. This can be used to improve fraud detection systems, protection customers as well as financial institutions.

Credit scoring: by looking at a larger amount of valuable data, a more accurate and comprehensive credit score can be assigned. This data can be traditional credit history supplemented with social media profiles, online behaviour and other purchases.

Generative AI can help financial institutions better identify and forecast market trends, risks and opportunities. By looking at historical data, generative AI algorithms can identify patterns and correlations to make predictions about market movements and trends. This is good both for customers looking to get a better return, as well as financial services who can manage capital more effectively and lower their risks.

While this all sounds like something we already have, the way generative AI algorithms work is fundamentally different. For example, you could have several agents running on historical data up to two months ago who all make predictions about the future market (relative to them). The best performers note why they did well, and then try again, and again, and again. This is done thousands of times until you have an overall best performer who then hopefully will fair well against current real world data.

Market forecasting: as described in the example above, generative AI can take historical data and try and predict market movements across stock prices, currency exchange rates and interest rates. This is key for institutions to better manage risks while aiming to get better returns on capital.

Risk management: like finding fraudulent activity in customer data, unsual patterns can be found and inspected for potential crashes or overvalued assets. Previous patterns can be investigated to better understand how a particular geopolitical event might move the markets.

Portfolio optimisation: with the ability to easily look back and understand past performance of different potential assets, this data can be used to create models that can better predict future performance. These models can then be used to dial up the returns while managing the risks.

Generating customized solutions

One of the most exciting aspects of generative AI is how it might change the way consumers interact with their accounts. With the latest tools, the chat interface has become the most natural way of interacting with these models. It's much more natural to talk to data and get answers, than to work through in excel or PDFs to get an understanding.

The technology itself is better too, so not only is there a different form factor for getting these insights, the insights themselves will be vastly better. By getting a better understanding of how certain pieces of data tie together more complete answers can be given.

Personalised investment portfolios: completely customised investments based on an individual's needs, risk tolerance, habits, goals and values.

Customised loan offers: no more being placed into super broad buckets for insurance and loans, generative AI can look at you much more as an individual than as part of a massive group. You'll be able to get better, more specific offers for credit and risk products.

Personalised financial advice: you'll be able to have your own financial advisor in your pocket, giving you advice on budgeting, savings, planning for retirement and whatever else your goals are. Because the advisor understands you through your data and your preferences, this will be completely tailored.

Implications for Traditional Financial Institutions

With any new disruptive technology, there are both challenges and opportunities. Due to how new the technology is, there are currently no regulatory frameworks to operate within. This is both a good and a bad thing, and governments are responding very differently. The UK is being open and trying to nurture the industry through a very light hand, while Italy has decided to shut things down while they figure out how to approach it. There is also an AI Act that is currently being proposed. While there is no regulation, corporations need to tread lightly about how they integrate this technology and what services they offer based on it. Data privacy and security are chief concerns and companies do not want to find themselves on the wrong side of the fence.

If the risks are adequately managed, the benefits are sizeable. Decision making can be improved by orders of magnitude both in the quality of the decision as well as the time taken. Personalisation of products and services can be vastly improved. Automation can lower the costs throughout the organisation, streamlining operations, reducing manual tasks and improving efficiencies.

There's no one size fits all approach to integrating generative AI into the organisation, but there are some key points that should be included. Collaboration with AI experts and technology providers to help develop and implement solutions for the specific needs will save time and get the right solution deployed. Managing expectations of the workforce will be key, with training to show how new tooling will make their jobs easier. A governance framework should be established to manage the associated risks, and if possible should be done in conversation with any relevant regulatory bodies. A holistic approach will de-risk any implementation while making sure the organisation is not left behind.

Conclusion

By understanding what generative AI is, you can quickly see how it can be applied into an organisation both for the employees as well as the customers. There are several applications for customers from better understanding to managing investment risks, and a huge amount of efficiency to be gained for the organisation itself.

There some risks associated, mainly linking to a changing regulatory landscape and data privacy concerns, but if these are adequately managed and a good implementation plan is created with experts and regulators generative AI technologies can be safely introduced in to large financial organisations.

The upside is unbounded: productivity for employees could skyrocket, costs for goods and services could be reduced by 10x or more, and completely net new capability can be introduced. The rewards far outweigh the risks, but the risks are definitely real. In time, organisations will need to have this capability to stay relevant so the question is to move first or to try and catch up later.