In my last blog post I looked at the pitfalls and possibilities inherent in today’s chatbots. In this one, I’ll address their challenges more specifically and how these challenges can be surmounted.
The Restrictions of Robotic Communication
Communicate with Siri or Cortana long enough, and you’ll quickly realise these chatbot technologies don’t truly understand what we’re saying. They work by picking up keywords. They still tend to stick to scripts and a defined set of questions, answers, and topics. As such, even a vaguely complex question can stump them.
But even if chatbots understand the question, the next hurdle is helping them access the information they need to answer it. Supplying this info can prove problematic — natural language processing needs to include contextual awareness so sentences can be registered, understood, and actioned accordingly in a timely manner.
Blending Bots and Humans
We are still at the stage where bots and humans need each other. While the industry continues to tout bots as a magic bullet for everything from app development to reducing headcount, companies are quietly employing human labour alongside to bear some of the strain as well as buy them time while developers work around the clock to implement AI solutions that can operate autonomously and scale as promised. The software is learning with every interaction, but we have yet to see whether or not these bots ever graduate to fully functioning, fully capable automatons.
Bots can respond to simple questions like balance inquiries, but for personalised and specific requests they must still deflect to human agents, which will remain true for the foreseeable future. In this respect, bots can be seen as acting as a front-line receptionist while filtering more specific customer inquiries elsewhere. As a tool in the customer engagement arsenal, bots, therefore, have value, but they don’t come close to living up to their promise yet. Bots alone will not save your broken customer connections. Of course, the more information our automated agents have to learn from, the better they’ll become. Brands that want to see bots taking centre stage in future need to begin by building rich archives of conversational data now.
Bring on the Bots…but Slowly
Right now, we’re looking at an immediate future were bots and human agents coexist in harmony, in service centres all over the world. But what matters in this hybrid scenario is both practical execution and a lowering of expectations, which many customers may not be willing to tolerate. Today’s bots are on the unintelligent and slow end of the spectrum, falling far short of consumer expectation. Waiting for a response from a human is, though never enjoyable, at least relatable. But waiting for software seems absurd — and infinitely more maddening if things go wrong.
It’s a delicate process to avoiding the disconnection that often occurs when machines try to mimic humans. Bots will need to queue and distribute requests seamlessly to their human colleagues and customers without the conversation feeling stilted or disconnected. We don’t expect traditional apps to care or empathise, but if software acts human, consumers are going to hold it to the same standard as dealing with a live person. With the bar set so high, anything less than perfect stands to be a huge let-down. Brands can’t avoid their responsibility for the customer experience by slapping on a “beta” label, in essence blaming technology.
In increasing numbers people are craving meaningful interaction, and using messaging as a playground for experimental AI could alienate consumers. Nobody wants this. Bots are essentially here to make life better, and we can only hope for even greater bot sophistication in the years to come. Ultimately this will free up more and more live agents to deal with the tricky issues that always need a human touch. In the meantime, brands would be wise to treat any consumer interactions as a precious opportunity for engagement that can build loyalty and confidence rather than rushing to hand them over to bots too soon.