During Module 5 of Squared Online we ask participants to complete a whitepaper on how future digital trends will impact a specific industry. The top 3 whitepapers are shared on this blog as a celebration of all their hard efforts!
Congratulations to Virgin Holidays team for this stellar piece of work on "AI Assisted Chatbots In The Finance Industry". Team members include: Nicole Gibson, Christopher Insall, James Libor, Jamie Marchant, James Burton-Lee, Joe Adams and Hayley Selby.
Chatbots, or intelligent virtual assistants, are a rising trend in many contemporary industries. Strictly defined as a “computer program designed to simulate conversation with human users,"1 they are distinct from common voice recognition functionality such as that on early mobile phones and the Xbox Kinect, which merely executed actions in response to vocal stimuli2. Covering both text-input and voice-input bots, chatbots arguably entered the public consciousness with the launch of Siri in 20113 .
The 2016 Gartner Hype cycle placed “natural-language question answering” firmly in the Trough of Disillusionment, around 2-5 years from mainstream adoption4 . While chatbots have not yet reached mainstream adoption, proliferation has already happened: every smartphone in the world contains a chatbot in the form of Apple’s Siri, Google’s Assistant and Microsoft’s Cortana (also found on all Windows 10 desktops)5.
While the financial industry is beginning to adopt chatbots, other industries are already doing so successfully. WeChat, China’s powerhouse messaging platform with over 800m users, is home to thousands of bots6 . A recent experimental Microsoft WeChat bot, XiaoIce (designed purely to listen to, converse with and entertain users) quickly grew to almost 40m users7.
The popularity of chatbots (in a more mature market) has been established: what has not is their utility. Analysts at Juniper Research suggests global business savings from chatbots could be $8bn a year by 2022, largely from reduction in customer-facing staff8 . To look at an individual example, the “DoNotPay” parking fine fighting bot has already saved its users $4m - but perhaps more crucially, thousands of hours of lawyers’ time9 . The creator of the bot is now attempting to replace other legal functions, such as applying for immigration status in the US10 . Consider the number of functions being replaced - legal advice, administration staff, immigration staff, web design of a form: it is impossible not to see the benefits of chatbots for finance. Financial advisers, bank staff, customer service, account applications, bank websites and apps: all could be replaced or supplemented with chatbots.
The Finance industry as a whole has gone through a technological revolution in recent times11 whereby financial organisations are being disrupted by start-ups challenging longstanding business practices12. Through these changes and the emergence of new technologies and even new currencies13 , customers have even more choice in how they interact with financial organisations and how financial organisations do business.
There are a few examples of start-ups within the finance industry who have pioneered chatbots, such as Chip16, a chatbot designed to help you save money as you spend, and Kasisto17, which is a whitelabel conversational AI available for banks to pick up and use off the shelf18. However, most of the companies are still only fulfilling basic service requests, and have yet to get sophisticated enough to replace the roles of financial advisor or stockbroker. This is a huge opportunity as the next step in AI is through cognitive conversation between a customer and a chatbot, designed to provide the customer with an effortless service when fulfilling requests currently requiring human intervention.
Strategy and Tactics
While chatbots are an exciting proposition for the industry, any disruptive shift comes with challenges, which are potentially heightened in the financial sector where consumer trust of banks is low and the stakes are high if it doesn’t work19. There has already been strong take-up of chatbots in customer service, where users value the ability to get access to finances in a way that works for customers, outweighed by risks and negative customer sentiment.
A parallel can be drawn to internet banking, which at first was feared by the mainstream consumer relative to the traditional in branch services, however given the vast improvements in service and accessibility, which benefit both customers and banks, internet banking has been the new normal for several years20. With messaging becoming ever more commonplace, and the millennial generation preferring to communicate via messaging rather than face to face (The Telegraph, 2012), with careful implementation it is only a matter of time before chatbots are the new normal.21
By 2020, the majority of financial institutions are likely to have chatbots in place. There are three key areas of consideration in the approach to implementation: Customer, employee and technology.
- Chatbots must deliver cognitive responses to make advice easy to understand and implement for the customer.
- They must be hassle-free, with notifications delivered via existing platforms such as Facebook Messenger (see, for example, American Express).
- Data security must be an absolute priority, and highlighted to the customer.
- Pledges must be implemented to always get a human to verify any complex financial situations.
- In addition, customers must have the option to speak to a human for additional support in the event of a chatbot failing to assist them.
- In the same way that internet banking has reduced the need for basic customer service by allowing customers to self-serve simple requests, chatbots should be embraced by bank employees as a way in which to replace the diagnostic stage of financial advice, and provide the financial advisors with the right information they need to support their product and customer.
- To obviate the need for complicated interfaces and extended training, store- and phoned-based employees could use chatbots to accomplish certain functions.
- For early adopters, new customers will be encouraged to join the bank due to the technological innovation in customer experience compared to rivals.
- Relevant, personalised CRM data on customers can be surfaced to customer service agents at the ideal time to deal with the current query.
- Investment in machine learning will be necessary to ensure that the correct insights are being drawn from data collected. This will mean the most relevant and helpful information is given to the customer.
- Investment in computer science would mean the most cognitive and human-like responses can be sent to the customer, which will provide further reassurance and trust when utilising chatbots.
- As customer apathy to chatbots is high, continuous rounds of testing must take place to ensure customers find chatbots beneficial and so improvements can be made to any issues.
- Investment in machine learning & computer science will be seen as an initial barrier, but this will be offset overtime as the chatbots deliver improved customer experiences and potentially reduce other business costs.
Benefits, risks and challenges
The benefits of banks and other financial institutions implementing chatbots for financial advice and customer services are many. Firstly, deploying chatbots will reduce overheads across many areas of the business, such as reducing the property portfolio and number of customer service staff. For example, the Royal Bank of Scotland was able to automate basic customer service queries, allowing staff to focus on more complex issues22. The converse of this is reduced employee loyalty: if staff see themselves as being replaced by chatbots and AI, they may seek alternative employment if they believe they are next23.
For customers, the rise of financial advice chatbots will democratise access to what was formerly a preserve of the moneyed elite. Robo-advisory platforms such as Nutmeg have already opened up long-term stock market investment to millions, but chatbots will provide a far more personalised and tailored service24. While Nutmeg and the like offer options and multiple levels of risk, they cannot replace an Independent Financial Adviser for a customer with specific goals: e.g., saving for a house, managing their personal budget, or saving to start a business.
The risks of such widespread automated financial advice are obvious: bad advice. If banks are providing chatbots, they may (as your bank’s mortgage advisor would) steer customers towards the bank’s own products. More maliciously, a chatbot may encourage a customer to get into debt or take out excessive credit, in order to increase interest payments, overdraft fees or late payment fees.
Furthermore, not all bad advice is malicious. While a human can understand subtext and intent, a chatbot may struggle. Thus the risk of misinterpretation comes to the fore - one only needs to consider Microsoft’s Tay, which only a few hours after deployment, spouted nothing but hate speech25. Misinterpretation of a customer’s wishes could leave a company open to lawsuits, if not properly designed and controlled. There could lead to a scenario where consumers are miss-sold or given poor quality advice on an industrial scale, as per the PPI scandal in the UK26.
Heavy regulation will need to be in place to deal with these potential issues. In the event that a chatbot is presenting a user with terms and conditions then the bot will need to recognise this scenario and be able to keep a record of the terms accepted by the user27. It’s important that any advice offered is informed, correct and highlights all associated risks. It has been reported that 70% of Facebook chatbots fail to understand the user so in an event that the bot is unable to answer then a disclaimer and/or potential human intervention needs to be considered. Highlighting risks and knowing when to pass over to a human coincides with being up to date with the latest regulatory changes, which can hit 185 a day for global banks. Since 2008, banks have paid $321bn in fines due to conduct breaches, so using auto-updated chatbots which intelligently adapt to the latest regulations could save banks large amounts28.
In addition chatbots within the financial sector are highly likely to have access to multiple levels of consumer PII (personally identifiable information). This highly sensitive information may be traceable within the transcript and so it is imperative that the highest levels of data protection protocols are followed, especially when the bot is hosted on third-party platforms such as facebook or Alexa. Chatbot security will be expected to adhere to the new GDPR regulatory changes due in May 2018 in the UK29.
Chatbots are only a component of a wider trend towards artificial intelligence entering the workstream. Until recently, AI has largely lived behind the scenes in applications such as high-frequency trading, voice recognition and video games30. With the advent of chatbots, billions of users will be directly interacting with AI on a regular basis.
Companies in the finance sector will need to get onboard with this trend for chatbots. A bank without a mobile app or responsive website in 2017 would lose customers as a result: it is likely that customers will feel the same in 2027 if chatbots become a new norm for human-computer interaction.
- For features, see: https://www.apple.com/uk/ios/siri/; https://assistant.google.com/; https://www.microsoft.com/en-gb/windows/cortana
- GDPR Readiness Report, Virgin Holidays, 12/06/17 (internal doc)
Find out more about Squared Online: you can give us a ring on +44 (0) 20 7173 5938, or download the brochure to read about the course and the Squared experience.