With all the innovation that has occurred in the lives of consumers because of artificial intelligence (AI), very little has been done to meaningfully change how people manage their personal finances. As a society, we will benefit from approaching personal finance in a fundamentally different way to motivate good long-term habits. Fear of bankruptcy, hope of having enough money for retirement, inability to plan to put the kids through college, counting on the stock market rising is for many people the current – but wrong – way to address these important issues. Hope is not a strategy.

We can now impact the way we interact with money, rooted in a deep understanding of the complex equation that is an individual’s financial situation, their goals, the economy, markets, tax rules and many other factors. A new model exists for how people can have an informed relationship with their finances, and take control of this important aspect of life. A new form of information and education can be economically adopted by the public at large, resulting in the acquisition of new habits. The challenge of economic illiteracy that plagues most of society – even our most educated citizens – can be addressed with AI. Today, only individuals with substantial net worth can access the services of professional financial advisors to approximate this level of service, but even those insights are not as comprehensive or tailored as what today’s computational and machine learning power can create. A human financial advisor cannot process the hundreds of financial decisions that their clients make, let alone layer on top of that the changes occurring in state and federal tax codes, changes in inflation, the stock market and many other variables – as well as the interplay among those variables. Only advanced computational technology that can process millions of data points per client can adequately digest this web of interrelated information – and make sense of it. Because each person is unique financially, there is no “rule of thumb” or general guidelines that can give an individual the right advice for them. It must be highly tailored to have value.

Because of the transformational changes in technology (including computational power, cloud storage, smartphone and interactive browser technology), we are on the cusp of using well-known AI techniques to build remarkably low-cost, 24/7, fiduciary AI-powered financial advice for the public. Some of the important AI methodologies include:

• Feedforward neural networks – modelled on how the human brain works. These are able to process interrelated information and understand relationships between many different variables which have complex relationships. Humans are limited in how many of these relationships we can analyse; however, we make decisions that mimic a feedforward neural network all the time – we understand intuitively that if we want to send our children to an expensive college we may not be able to buy the house we want, and retirement may be delayed. What we are unable to do, given the limitations of the human brain, is to layer in other continuously changing factors, such as inflation rates, tax rates and other expense changes, to quickly determine the full range of choices at our disposal and the most optimal combination of these choices for us. These techniques are also able to consider things like short-term vs long-term trade-offs, which are of course inherent in financial decision-making. In short, the feedforward neural network is a highly sophisticated “brain”, which can process and make optimal decisions across a nearly infinite number of variables and their relationships with one another.
• Reinforcement learning – the next layer that can be added in AI methods to truly provide individualistic and specific advice unique to each user. Reinforcement learning takes the details of one individual’s behaviour and feeds that into machine learning algorithms – creating a growing insight into the objectives and behavioural patterns of that individual user. Rooted in behavioural psychology, this aspect of machine learning is very important in moving away from generic advice and into more complex problem solving with unique outcomes.
• Collaborative filtering – a term that may not be familiar, however, we are all quite accustomed to how it works in our lives, such as books being recommended on Amazon in the “if you liked this book, you may also like these” format. It is the ability to make predictions about the interests of a user by collecting the preferences of many users. This becomes relevant in many areas of life, including things like managing expenses, planning for family vacations or determining objectives for investing. While extrapolating preferences from a large sample of general information is a good start, it is not tailored enough for appropriate financial planning. The financial advice each person needs must be based on their own unique data and circumstances.
• Proliferation of interface AI for speech and language – simplistic interface AI is a familiar part of the user experience in many aspects of everyday life. Consider the verbal phone “bot” that allows you to speak your information as you navigate through an automated menu of options when you call your bank, or other large company. While most AI bots today seek to understand and communicate simple instructions like “What is my balance” or “What’s my spending on coffee?”, we can now answer more complex questions, like “Am I saving enough?”, “Can I afford to send my child to an Ivy League University?”, or “Can I retire at 60 instead of 65?” Linking interface AI with computational AI can both understand the question and provide complex answers with detailed reasoning, at your fingertips.

Combining these interfaces with an AI financial advisor is a ground-breaking and important advance, because ultimately people want to receive information in a way that is familiar – as if they were speaking with a human being. Leveraging tools like Amazon’s Alexa or Siri adds a dimension of ease – and is also an important advancement for users who are visually impaired or have challenges with dexterity. Fundamentally, when people can no longer distinguish between speaking with a human or with a machine, it becomes easier for the machine to be leveraged inexpensively and with scale – and across many languages. This also allows the machine to proactively reach out when there are issues, changing the dynamic from a person having to sift through their financial accounts to ascertain whether they need to worry about something, versus a new paradigm where your computer, phone or Alexa can tell you “hey, take a look at your bank account, it looks like you may have double paid a bill”, or “those new tax laws that were just passed by Congress will have a negative impact on your retirement savings, unless you consider retiring in a more tax-friendly jurisdiction, you will run out of money sooner than anticipated”. This is the sort of “nudging” that we really need from our automated AI-driven financial advisor, and we are on the verge of this being a reality.

Advances in computing power are one of the main drivers to leverage data and to change how we incorporate AI into our lives. The advent of the cloud has enabled inexpensive and large-volume computations. It has made enterprise-grade infrastructure security standards and military-grade encryption more easily accessible, enabling companies of all sizes to have substantially higher levels of security than most people experience with their financial institutions today. It is worth noting that many large corporations must redo decades of relatively open security, while newer companies can build from the start using sophisticated encryption techniques and other methods that greatly enhance security.

How does regulation keep up with these technological changes? What many regulators are beginning to understand is that advanced technologies are the best hope of delivering unbiased, appropriate and fiduciary advice – such as what is required by the US Department of Labour (DOL) fiduciary ruling, a rule which expands the “investment advice under the fiduciary” definition under the Employee Retirement Income Security Act. Unlike humans who – even with the best of intentions – come with biases, an AI financial advisor can be programmed to be a fiduciary and must only give unbiased advice by design. This is a huge protection for the consumers of financial services, and one of the most important dimensions of how technology will impact this sector. We are not far from the day when advice given by a human advisor will be compared with what an AI-based advisor provides, and it becomes increasingly complex to justify the cost of human advice. This is like investment management, where studies have shown that over the last 20 years up to 92% of actively managed mutual funds underperform an index to which they are benchmarked.

The future lies in intelligently harnessing the power of these tools to provide AI-based, user-friendly solutions, especially for complex and expensive problems like holistic financial advice. Given the acute need for reasonably priced approaches to solve the questions that people have about their finances, it is AI and only AI that can provide this. The future of financial advice is personalized AI, and it is the power of imagination that makes the impossible happen.