As the chatbot world continues to expand, we in the tech industry face some critical questions. How will we measure success? And how do we ensure our approach to development drives successful outcomes?
It’s not simply enough to set a standard goal, like developing a chatbot that passes the Turing test. Instead, we must seek to create chatbots that align tightly with specific business goals and objectives — expressly designed to perform tasks and actions that solve real problems. They should be charged with the mission of moving needles in measurable ways and transforming key performance indicators that ultimately impact business performance. And to be truly effective, chatbots should strive to be agile, scalable, and omni-channel in nature.
This, of course, is much easier said than done. Many companies are able to tackle smaller, standalone chatbot projects — say, for Facebook Messenger or the web. However, the majority of us don’t possess the resources or manpower to develop fully comprehensive implementations with game-changing customer experience potential.
A big part of the problem lies in how we tackle development. Typical legacy chatbots are strictly rules-based and require a great deal of maintenance and manual work, such as the tedious task of data-labelling. The process often looks something like this: Write the code, launch the bot, add more labelled data, modify the rules for additional use cases, rinse, and repeat. This needs to be performed on a continual basis in order for the chatbot to deliver on the promise of an “automated” experience. And since everything has to be hand-coded and managed directly by developers, there’s little opportunity for real-time intervention by the customer experience (CX) leader or any other relevant business user.
These time-consuming, resource-intensive development practices are what hinder smaller companies from achieving effective implementation — not to mention the cutthroat competition for the limited pool of engineering talent not already claimed by Fortune 500 giants. So what choice do these organizations have? They either go the route of pouring all their engineering resources into chatbot maintenance or they outsource their needs to a third party. Neither scenario is ideal, because they’re either being terribly inefficient or they’ve lost considerable control in designing the kind of experience they want their chatbot to provide.
If the goal is to create impactful chatbots that deliver superior user experiences, we need to rethink the role of AI development and how it can best serve this objective. We should create smarter, more flexible AI that enables rapid application delivery with minimal hand-coding and can use any combination of labelled or unlabelled data. We must conceive of a development approach that truly optimizes human-in-the-loop computing, maximizing the proverbial synergy between man and machine.
Doing this will shift the ownership of the chatbot away from engineering and place it in the hands of the domain expert who can truly maximize its business potential: the CX leader, for example. You can think of it as a way to sidestep a huge chunk of the application development and delivery process and hand the keys over to the business user instead. If we do this, we won’t just create better chatbots, we’ll free up our engineers so they can stop wasting time and energy on rote maintenance and focus on bigger and better things.
Why should we place the power in the hands of the CX leader? Simply put, AI works best within chatbots if it can infuse them with a human sensibility. Chatbots are able to observe, learn, and self-improve best when they have a real human context at their disposal and can integrate human knowledge, such as best practice and work flows. This means that the best person for creating a customer service chatbot isn’t just the engineer with the right technical credentials but one who has an intimate understanding of and personal experience with direct customer engagement.
A better approach to AI chatbot development will create a foundation for CX leaders to do what they do best — create happy customers. It will provide a versatile deep learning base that empowers CX leaders to build tasks and complex work flows without ever having to touch a line of code. This will enable them to further hone in on the key performance indicators that will bring the customer experience to the next level. This may involve A/B testing to compare the metrics for a virtual assistant against live agents, interactive voice response systems, or self-service FAQ and measuring everything from efficiency metrics and deflection rates to net promoter scores and customer satisfaction ratings.
Perhaps most importantly, this development strategy will allow companies of all sizes to create scalable, omni-channel chatbots and maintain and evolve them in an affordable, efficient way. Rather than hiring an engineer to build the stage, hire the musicians, craft the instruments, arrange the seating, compose the music, and conduct every time we want a little music, we can allow the CX expert to simply take their place at the conductor’s podium and orchestrate the entire experience from the get-go.
VentureBeatVentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
- up-to-date information on the subjects of interest to you
- our newsletters
- gated thought-leader content and discounted access to our prized events, such as Transform
- networking features, and more