The financial services sector is beginning to appreciate the sheer scope and scale of how artificial intelligence (AI) could impact everything the sector does. However, AI is still engendering genuine stupidity from those who are writing the technology off, or who fail to appreciate the scale of change and human impact it could bring about. So, how is AI being applied in financial services today, and how could it evolve over time?

AI is already being used in applications such as customer service, trading, and targeted marketing. Internally, applications include automated report writing, reconciliations, risk management, and fraud detection. Successive waves of AI FinTech start-ups are moving even faster with the emergence of AI-powered hedge funds, robo-advice, multi-factor stock analysis, personal finance managers, copy trading, and numerous applications that seek to streamline traditionally slow and costly processes in areas like book building, invoice financing, and personal insurance.

So, what is this all-powerful technology which many believe could transform society and enable whole new industries? The core premise of AI is to create computer algorithms that replicate aspects of human intelligence including planning, scheduling, problem solving, reasoning, inference, image recognition, language interpretation, and speech. There are several sub-fields under the AI banner including expert or rule-based systems, robotic process automation, chatbots, machine learning, neural networks, speech recognition, natural language processing, robotics, and computer vision.

A number of factors have contributed to its progress, including the emergence of sufficiently big data sets for AI programs to learn from, faster hardware and communications, cloud connectivity, and more efficient AI algorithms. Perhaps the most important factor is the multi-billion dollar investments being made by tech giants like Google, Amazon, Alibaba, and Baidu.

The technology is already evident in financial services with a diverse range of internal applications: real-time detection of fraudulent banking and credit card activity – instantly preventing situations that humans could take weeks to rectify. Another increasing area is the automatic generation of commentaries and reports from underlying data using tools like Quill. Many institutions are deploying chatbots to converse directly with customers. Examples include Cora – Natwest’s ‘Digital Human’ and Nordnet’s Amelia. Mastercard have created a virtual sales assistant that captures the learning and best practices from the human sales force, freeing staff up to handle more complex tasks.

AI’s capabilities are evolving rapidly with mind-blowing breakthroughs being announced daily

Leading adopters such as Blackrock have been deploying AI for some time, for example replacing humans with algorithmic fund management. They have also applied AI for analysis tasks including company reports and announcements, producing earnings statements, and evaluation of retailer locations. Others are using AI to manage the breadth marketing content and channels, to upsell customers (eg Prudential), and in influencer marketing (Influential). One interesting opportunity AI is being targeted at is in trying to reach and build relationships with the younger members of wealthy families who stand to inherit the responsibility for managing US$30+ trillion assets globally.

AI adoption is accelerating, led in many cases by FinTech start-ups. These include new digital banks (eg Monzo, N26), provision of personalised financial advice according to life stage, assets, and needs (Pefin), undertaking multifactorial research and analysis of individual stocks (Capital Cube), and robotrading (Collective2). Others are bypassing the banks to crowdfund corporate debt from retail investors (WiseAlpha), offering higher interest rates to savers via peer to peer lending (Lending Crowd, Zopa), allowing individuals to copy the best cryptocurrency traders (Covesting), and creating algorithmic hedge funds (Numerai).

AI’s capabilities are evolving rapidly with mind-blowing breakthroughs being announced daily. For example, the capability of machine learning and deep learning systems is evolving to the point where they can learn about multiple fields of knowledge without forgetting anything. A team from Alibaba and Microsoft recently outperformed the best humans and competitors from other technology firms in a Stanford University reading comprehension exercise designed to test the interpretive capabilities of AI systems. Researchers in Japan have developed an AI ‘mind reading’ neural network that can analyse brainwaves to produce rough approximations of the images being viewed by the human test subject.

Massive investments are being poured into accelerating AI development by countries and companies across the globe – with China seeking world dominance by 2030. While it is hard to predict exactly how this fast-moving technology will play out over the next 20 years, we can suggest a possible timeline.

The widespread use of autonomous vehicles could see the emergence of self-owning vehicles

In the period from now to 2020, we can expect a range of advancements, with AI being adopted by all financial services firms either through in-house developments or ‘AI as a service’ applications. Uses might include real-time translation enabling us to serve clients around the globe without being fluent in their language, smart machines, sensors, and objects capable of carrying out basic transactions, and self-editing software that will allow insurance risk algorithms to update in the light of changing social behaviours.

Perhaps more of a mental stretch is the notion of fully automated smart financial corporations (decentralized autonomous organizations) with no employees. Monitoring of fraud and insider trading could evolve to track fund managers and personal advisors.

Tracking systems would monitor transactional behaviour, personal spending, and social media posts to spot potential fraud, abuses of client funds, market manipulation, insider trading, and money laundering.

Comparison of our personal data with millions of others would enable automated intelligent finance advisors to recommend cheaper alternatives for goods and services we buy regularly and even to aggregate our purchases to secure bulk discounts. Users could authorize the system to monitor constantly for the best options and switch our savings, insurances, and purchases accordingly. Conversely, marketing and customer interaction may get a little harder when customers start to use intelligent agents (think Siri++++) to manage their lives and guard their personal data, sharing only the absolute minimum with those who provide them with financial products and services.

Looking ahead to 2025, it seems likely that 70-90% of customer interactions across financial services product lines could be managed by AI, with new product and service development increasingly undertaken and tested by AI. The technology should be sufficiently advanced that banks and insurance companies could offer platforms which allow customers to define completely customised products and services to meet their individual needs – with other customers investing in the provision of these offerings. The widespread use of autonomous vehicles could see the emergence of self-owning vehicles, the end of personal ownership, pooled insurance across the hive of intelligent vehicles on the road at any one time, and vehicles bidding for other cars to move out of the way because the passenger is in a hurry.

There is a near limitless range of potential applications for AI across financial services

New opportunities might arise to integrate personal financial management with our spending on clothing, entertainment, travel, and dining. The systems could be empowered to trade unused airmiles and store points, negotiate discounts, offer our car into a sharing pool when unused, accept paid adverts to our social networks, and rent out bedrooms (eg to sites like AirBnB), cupboard space (temporary storage), and driveways (parking). Cash surpluses on our account could also be invested and switched on a moment by moment basis, guided by our preferred risk profile.

Scanning the far horizon out to 2035, the financial world could be transformed completely through the combination of AI and blockchain – the underlying secure distributed ledger technology for most cryptocurrencies. Indeed, the concept of money could evolve into electronic tokens with far more types of assets tradeable within the one ‘currency’. For example, we might earn tokens from our employment, as rewards from retailers and airlines, and as micro-credits for completing workplace training or school learning tasks.

Instead of simply liking a track from a musician, we could now make a micro-payment to them with a fraction of a token. This evolution from cash and cryptocurrencies towards a universal means of exchange could mean the end of cash and foreign exchange markets. New opportunities will emerge to manage such platforms, invest the tokens, and to create meaningful equivalence across all the different asset classes covered by such tokens.

Clearly, there is a near limitless range of potential applications for AI across financial services. From streamlining activities and reducing costs through to new product development and customer service – the opportunities and benefits are starting to emerge. From a customer perspective, AI offers the potential to help make our finances go further and unlock the value in assets that have not been tradeable in the past. From where we stand today, it is reasonable to assume that AI will become core to the future functioning of the financial ecosystem.

Rohit Talwar is the editor and co-author of The Future of Business, Beyond Genuine Stupidity – Ensuring AI Serves Humanity, and The Future Reinvented – Reimagining, Life, Society, and Business. See more at