How far are you willing to trust chatbots?

Artificial intelligence (AI) looks set to continue revolutionising the banking industry well into 2017, with many banks utilising AI software to increase their security measures and improve efficiency. Recently, these have been focused on biometric measures, including voice and facial recognition systems, along with customer service tools. As I have discussed in a previous blog post, AI in the financial world opens up the ability to analyse huge data sets, extrapolate trends and then learn from past experiences to provide better solutions. As such, the AI can make self-service better and, in the case of chatbots, can provide an enhanced and faster experience for customers.

Efficiency is everything

As customers continue to demand more from banks, chatbots are proving to be a great opportunity. Banks and customers alike can spend less time solving simple problems, such as basic transactions or overdraft changes. This frees up customer service agents to handle more complex enquiries, which are often higher value. For customers, the promise of chatbots can mean the end of battling with call centre menus or waiting for an agent to become free. There’s little more frustrating than sitting in a hold queue for a call centre, right?

Can chatbots have a human touch?

A recent report by Mindshare UK found that 63% of people would consider messaging an online chatbot when they need to get in touch with a company. Not surprisingly, 79% of respondents wanted to know whether a human could step in if the chatbot was unable to help or confused by a more complicated query.

These results show the clear opportunity there is for chatbots in the market, but they’re far from perfect at the moment. With developers battling to create the most intuitive system that learns from its past experiences and conversations, there’s a lot of potential for huge advances in chatbot capability over the coming months and years.

Language barriers are still a real problem. Although language rules can be programmed, often humans do not follow these rules when communicating. This leads to frustration for the person using the service, which of course causes even more confusion for the chatbot – think Microsoft’s Tay bot, which ended up spewing insults at users!

To combat this, chatbots need to be advanced enough to stop common expressions being misunderstood. Having said this, they are improving. I have found myself being unsure as to whether the customer service I’ve received has been delivered by a human or a chatbot – a clear sign that they’re starting to get it right. 

With chatbots, the tone of voice they use can make all the difference and an inflexible service can cause the user to feel de-humanised. Yet, on the other end of the spectrum, consumers have reported they can come across as ‘creepy’ if too friendly. An experiment was done using IBM Watson, in which two chatbots were built to have the same function with differing personalities. Users found the ‘chattier’ bot almost patronising, which put off users.

The importance of trust

Trust is a huge talking point in the financial industry as banks focus on rebuilding their reputations, and it’s no different when it comes to chatbots. Research has found that people are less willing to forgive mistakes made by an AI instead of a person, so banks need to be clear about what their chatbot can help with and rigorously test before launching.

However, strangely enough, a recent survey found that 25% of respondents are more willing to give sensitive information to a chatbot rather than a person, as they find them more reliable and more accurate in their responses.

Overcoming the barriers

Some critics say that the capabilities of artificial intelligence technology will never be intuitive enough to understand the needs of humans, and certainly won’t be able to deal with the complexity of financial decisions.

Looking to the future

One way to battle critique is to provide education on the chatbot technology and its possible uses to make it seem like less of a challenge to use the software and help to move the usage of chatbots inside the user’s comfort zone. This is likely to be the biggest challenge for many banks looking to implement chatbots.

If chatbots are able to overcome these teething problems, they could transform customer service in banks, increasing efficiency and cut costs spent on employing thousands to man call centres.

I think the biggest threat to their success is communication problems, capitalising on machine learning, to make sure the technology can intelligently take on board requests and understand the context of the questions. An option to improve the responses of chatbots could be to employ copywriters alongside programmers for more natural predetermined answers.

In my opinion, the best way forward is to slowly incorporate this technology into your everyday offering, rigorously testing their communication effectiveness to make sure they help rather than hinder users. This can be done by keeping open human communication lines and making the chatbots a secondary option. Educating the public on the technology, as I previously mentioned, to make them aware of the opportunities and limitations is also key. Finally, listen to your customers, encourage them to provide feedback and, most importantly, take it on board.

What do you think of the future of chatbots? Will they continue to develop and prove themselves a viable option for banking customer service?