Insight

Chatbots: Reality vs Perception

Ross Miles - 08.05.2018



“If you can’t tell if something is real, does it matter?” asked Angela (a ‘host’, or android if you’d rather) of Logan (a human) in episode two, season two of HBO’s hit TV show, Westworld. Set during an unspecified time period in the future, the human race has developed artificial intelligence (AI) to a level where it is indistinguishable from our own, and when installed in a life-like humanoid body, can be passed off as being a living, breathing human being. This scene in particular drives home this point, a cleverly devised sales pitch intended solely to shock Logan into becoming an investor in the company producing these ‘hosts’.

Therein lies the crux of AI in today’s society – a drive for artificial intelligence to pass the Turing Test. The test, theorised by one of the greatest Britons to ever live, Alan Turing, is a test of a machine’s ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human. The test itself is a text-only conversation between two-parties with an evaluator trying to determine which one is not human. To pass the test, the evaluator must not be able to identify that the machine is, in fact a machine.

Now with that in mind, consider the Live Help functionality often found on eCommerce sites today. The ability to be conversational and available 24/7 with customers has been shown to be financially beneficial to a business, and is also leading to higher levels of customer satisfaction. In fact, live chat has the highest satisfaction levels for any customer service channel, at 73%, compared with 61% for email and 44% for phone. As a user, when you have interacted with a live chat functionality online, were you aware you may have been talking to a chatbot? Does that change your perception of that interaction, and does it even matter?

As shoppers and internet users we’re growing more and more accustomed to using chatbots, with 53% of customers in saying they would prefer to use online chat before calling a company for support. Unsurprisingly, this is a trend that is stronger in millennials than baby boomers (millennials are 20% more likely to prefer live chat to have their basic customer support questions answered), because as a society we’re being conditioned to value speed and convenience. From microwave meals to contactless payments and self-service checkouts, Uber and Netflix, the message is clear to business owners – convenience sells, or rather, retains your customers. And as the old adage goes, it costs five times more to acquire a customer than to retain one…

So, chatbots. A recent IBM report suggested that 80% of customer service requests could be dealt with by a chatbot, with further studies suggesting chatbots can cut customer service costs by 30%, doubling down the commercial benefit of using chatbot technology by not only increasing sales, but also reducing outgoings. But do these chatbots need to be Turing-test level smart to adequately service customers? Since we’ve yet to see any machine pass the infamous test, the answer is clearly “no”, but a disjointed and unsatisfactory interaction with a chatbot clearly won’t meet the criteria of speed and convenience that users are looking for.

Equally so, just having a chatbot for the sake of having one doesn’t help anyone. That’s not to say there aren’t many uses for artificial intelligence in chatbot form that can enrich our lives, from a localised weather forecasting bot, a scheduling bot or even a new digital best friend, but jumping on the bandwagon and getting a chatbot of your own won’t suddenly increase your business’ profitability.