Recently, Craig Borowski, Customer Service Market Analyst for the online technology consultancy Software Advice, released a new report which studied if better customer service is the “necessity” that AI needs to be relevant.
We had the opportunity to speak with Mr. Borowski to learn a little more about what his research uncovered.
What have been some of the significant AI developments over the last few years?
AI implementations are built upon three components, and all three components have developed significantly in recent years.
The first is the hardware component and includes the CPUs, GPUs and other custom-designed chips.
The second component is the software, the actual code and algorithms that manipulate the data. Much of the software that’s relevant to AI comes from the field of machine learning, and relies on newly developed neural networks and deep learning methods.
Lastly, there’s the data. AI systems need to be trained and refined with large data sets. Large data sets are now more available than ever before, both in the private and public sectors, giving a wider variety and larger number of people the ability to experiment with AI.
Has your research found that most customers prefer to interact with a chatbot as opposed to traditional live customer service?
Great question, with a surprising answer – customers don’t really care! Their preference for a fast, accurate service resolution supersedes all other preferences, within reason of course.
If a chatbot and a live agent both deliver the same answer, but the live agent takes longer to provide it, then the chatbot will be the stronger preference for most customers.
Is it easier for customers to detect when they’re speaking to a chatbot instead of a live agent?
Yes, for now anyway. Customers are usually tipped off by the language the chatbots are programmed to use. It often seems very scripted and formulaic, and people are very good at picking up on this.
The most natural feeling chatbot interactions are those that use a chatbot for the initial one or two questions. Then after determining the nature of the service issue, seamlessly transfer the conversation to the appropriate live agent.
In these implementations, the chatbot functions like an IVR (for phone calls) without all the frustration consumers have (rightfully) associated with IVRs.
Should SMBs consider AI implementation? If so, why?
There are compelling reasons for SMBs to consider AI and chatbot implementations.
There is a quickly growing and very affordable selection of chatbots that can be used to assist customers at various points along the customer journey, like: researching or asking about a product on social media; asking customer service questions on a company’s support page. Chatbots offer a very attractive, scalable solution to managing these aspects of a customer relationship.
But, SMBs do need to carefully consider the value proposition and potential ROI for their particular situation.
Chatbots are great at handling large volumes of relatively simple, or relatively similar questions. If there’s a great deal of variety or complexity in the questions an SMB gets asked, then it’s less likely that chatbots currently offered will work without extensive customization.
That said, enterprises face the same challenges. Perhaps most interesting is the fact that AI is quickly becoming an IT disruptor, while simultaneously, it’s within reach of businesses of all sizes.
What are some of the trends you see AI software benefiting SMBs in the future?
One side that doesn’t get discussed enough is the potential for AI to transform SMBs internally, with employee-facing chatbots.
Imagine, for example, a small software developer with a staff of mostly engineers. Bug tracking is central to the process of software development and creates a lot of monotonous work for the engineers, entering new bugs, describing them accurately, checking that they haven’t already been reported.
One trend that has huge potential here is the implementation of chatbots and virtual assistants to help employees reduce the amount of time spent on routine and repetitive tasks. Since the workflow processes involved with bug tracking are consistent and predictable, they’re a perfect example of the type of work that will soon be handed off to AI applications like chatbots.