We specialise in building bespoke AI solutions for companies by delicately stitching together various AI systems. This means that the use cases for our solutions are extremely varied, but a lot of the use cases that we encounter are finding ways to standardise human behaviour to improve the customer experience.
For example, customer service agents in bank branches answer the same question in completely different ways, which has a direct impact on customer satisfaction. I’ve found it particularly interesting to discover that the longer a person has been in a job, the higher their confidence is that they’re providing the correct answer, but they have a lower accuracy rate because they’ve formulated their own answer in their head that they’ve repeated over time, and no longer check centralised information to review whether their version of the answer is correct, or has been updated since they started.
We developed an AI solution that combines Natural Language Processing and cognitive decisioning to provide the customer service agents with the correct answer from the centralised information system, thus standardising the response that all customers receive and removing customer dissatisfaction due to inconsistency of responses from multiple agents. It also enables customer service agents to focus on delivering high quality customer experience, rather than trying to rattle off an answer that they’ve memorised.
The biggest problem that I’ve encountered when delivering AI solutions is the inconsistency of human behaviour.
I went to the the COGX Artificial Intelligence Conference last week and, unsurprisingly, the discussions tended to veer towards discussing driverless cars and what this means for the future. There was a lot of debate surrounding responsibility, insurance and legal parameters, and how driverless cars will dramatically improve traffic and reduce the number of accidents on roads by removing human behaviour from the driving experience. However, for me, one of the key points to consider when talking about driverless cars is that they will only work if there are only driverless cars on the road. We know that humans can be reckless, which is why speeding holds such an appeal and why drink driving is an issue, but we also don’t react quickly enough or necessarily in the best way when a situation occurs. So the only way that driverless cars can have the positive impact that keeps being promoted is if they are on all of the roads, and we eliminate human driving and the behaviours that result in traffic jams and accidents. Which in turn begs the question – will we ever be at a point where driverless cars have monopoly on the roads that will enable them to be a success?
My principal concern with autonomous vehicles is that whether they are successful or a failure, it will have a direct impact on the public’s view of Artificial Intelligence, as this is the most accessible form of AI, which is why, whilst it frustrates me, I can understand that AI conferences and talks are dominated by the topic. Yet what we’re missing is the potential that current AI systems offer in terms of standardising and therefore improving customer service. We’ve all encountered a bad customer experience because we were provided with incorrect information, or the customer service representative has stuck rigidly to a script, however, the AI solutions that we’re delivering informs human behaviour to provide businesses with the opportunity to greatly improve information management and customer experience.
So although human behaviour may be the biggest challenge to adopting some AI technologies, such as autonomous cars, I’d actually like to flip this statement on it’s head. We have seen AI solutions which improve human behaviour by taking the drudgery out of everyday tasks and enable individuals, such as customer service agents, to focus on the human element of their work, allowing them to build more meaningful relationships, resulting in improved customer satisfaction.