- 0207 078 9830
- 01252 411 111
- 01483 339 993
- 0151 2030 200
- 0161 8700 700
- 0191 3000 666
Marks & Spencer recently announced their roll-out of an AI (artificial intelligence) communication system. It is designed to automatically understand and route voice calls to appropriate departments. Although they claim to have achieved over 90% accuracy, robots are not necessarily the direction we should to head towards.
Manu Tyagi from Infosys Consulting states that “consumers still prefer the reassurance of human interaction – and this should not be ignored.” We agree.
These studies highlight the importance for organisations to embrace remote working. It will soon be the only way to attract the best talent and keep them.
As one of our clients St John’s Medical Centre found, many organisations lean on ageing communication systems that cannot:
- Cope with current communication methods
- Be upgraded to suit changing requirements
- Support multiple offices through one system
Traditional communications solutions are therefore quickly becoming out of date. This creates more work and thus places a strain on staff.
Are robots the solution?
The first thing to remember is that true AI does not exist – we do not yet have a machine that can think for itself.
But more importantly, the human element cannot simply be ignored, especially in certain sectors. This was illustrated by Durham Police, who launched an AI system to decide whether or not a suspect should be held in custody. This however, led to a major concern.
The system assessed the risk level of an offender and made a decision on that basis. But, this created the risk of incorrect decisions since it was designed to err on the side of caution. Also, the risk assessment was only based on Durham’s police data – if an offender committed a violent crime outside of Durham, it was not taken into account.
So, although the system was rolled out, it still required a police officer’s input.
We therefore need to reconsider whether we can rely solely on technology to make decisions. In some situations, say in financial establishments, incorrect decisions can lead to disastrous consequences.
Let’s think about AI differently
We are on the verge of a digital revolution. Companies are increasingly looking for technology-based solutions with many turning to AI.
In some situations, a fully-automated approach can work. But when it comes to customer service, human interaction is still needed – “Humans will always be critical when it counts — from stepping in when bots can no longer handle the query to providing a real voice or friendly face to customers.” (Nick Ismail, Information Age).
Given the range of communication channels that customers have access to, it is tempting (and wise) to take an omnichannel approach to customer engagement. This is, finding solutions that can cope with the multiple sources of data, and streamline them into one big picture.
One of the ways that companies like Facebook and Mastercard are using AI to enhance customer experience, is through chatbots. This is where software can interpret a customer’s query – regardless of whether it is through voice or text – and deliver an appropriate response.
Although bots can handle simple scenarios like appointment setting and troubleshooting, there will always be situations that cannot be planned for. Ones that do not follow a specific set of rules.
In such situations, a blended approach of bots working alongside humans should be taken.
This approach can help organisations utilise technology to enhance service delivery and improve efficiency levels. In a contact centre for example, bots could be used to handle simple queries which would free up humans to deal with the more complex ones.
This interesting discussion between Paul Daugherty (Accenture) and Greg Williams (WIRED UK) during the DLD Conference 2018, explores how organisations are deploying AI and how machines and humans can work side by side.
How Microsoft became a racist troll on Twitter
Chatbots are traditionally divided into two groups:
- Ones that are based on a set of rules.
- Ones that can engage in ‘smarter learning’ from examples
Focusing on the latter, there has been a lot of discussion around how chatbots can become ‘smarter’ by learning from human interaction. The aim is to help them develop human nuances to enhance the scenarios they can be used in. But as Microsoft found, this doesn’t always work.
In 2016, they released a chatbot called ‘Tay’ that was given its own Twitter account and allowed to interact with the public. They announced that “The more you chat with Tay the smarter she gets.” But as people attempted to prank the robot, she started sending offensive tweets. Tay was shut down within 24 hours.
So, by looking at organisations who have used AI successfully and unsuccessfully, we can see that the best results are derived from a combined approach. After all, if a customer uses a chatbot that issues incorrect advice, they will conclude that it is unreliable and never use it again.
Consumers expect high quality of customer service. And if they don’t get it, they will give their custom to someone else. So, when it comes to using technology to enhance the customer experience, it is important to get it right.
Taking a blended approach
Given that a machine’s decision can only be as accurate as the data it has access to, we must think about AI differently. It shouldn’t be something that replaces humans. It should instead be something that works alongside them, to help them be more effective.
Ann Macdonald, a science writer for Diagnostics, discovered that in many areas of healthcare, “AI can already outperform the human.” However, with less common conditions “AI is far less affective [and] in almost all example use cases, a ‘human in the loop’ philosophy produces the best results.”
By recognising the importance of a blended approach, we delivered a futureproof technology-based solution for St John’s Medical Centre – without removing the human element of customer service.
Penny Borrow, Practice Manager found that their new telephone system “was future-proof, flexible and more than met our specification, particularly in terms of a call routing system that would improve the patient experience.”