The AI-Improved CX

In 1990, a fellow chess player and I attended one of Gary Kasparov’s several matches in New York City. Kasparov had become the youngest ever World Chess Champion in 1985, at the age of 22. In 1997, Kasparov also became the first World Chess Champion to be defeated by a computer, losing to IBM’s Deep Blue in a stunning six-game match. Computers have become so good at chess that phones are now banned at chess tournaments, and players are very carefully monitored for any sort of digital activity or connection.

Today, however, there is a new kind of chess becoming competitively dominant. Called “Advanced Chess,” or more colloquially “Centaur Chess,” it involves teaming human beings with computers, and it has been actively supported by Kasparov. It turns out that humans teamed with computers play better chess than either humans or computers alone can play (for now, anyway).

Many disciplines are finding similar models, from radiology and medicine to securities trading, and managing the individual, human-to-human customer experience is a discipline that is beginning to benefit from such a hybrid approach, as well. This is particularly true when it comes to helping companies manage their real-time interactions with customers – mostly involving phone and chat agents.

Of course, we’ve all had the experience of not being able to make ourselves understood to a chatbot, or coming finally to a dead end after toiling dutifully through the IVR. And the website GetHuman.com still does a roaring business showing human customers how to connect more quickly with human agents at whatever company or organization they try to penetrate. But a 2017 Forrester report listed four specific ways that AI-animated chatbots could be usefully teamed with human agents to improve the customer experience. And these ways parallel the manner in which a group of human chess players would team with computers to devise better and better games:

1. Chatbots for agents, rather than for customers: Let the human agents consult AI guides and chatbots in the background, even while the agents themselves are in live with conversations with customers. (This would be a straightforward analogy to a team of chess playing humans having one or more computer-chess programs open, in order to sample a variety of different potential moves, and to identify any potential mistakes or weaknesses in advance.)
2. “Human in the loop” AI: In this case the chatbot would suggest the right answers to human agents, with machine learning algorithms continually improving those answers, based on an agent’s actual words. According to Forrester, the airline KLM achieved a 50% reduction in handle time with this model, as their chatbots suggested answers for the more routine issues while agents took over when it came to more complex issues. (Think of a chess team consulting the computer to see a number of the computer’s own “most likely” next best moves, before choosing a move themselves, which might or might not be based on a more creative or surprising strategy.)
3. Front-end chatbot: This would expedite the human-to-human interaction by gathering routine information and vital (but predictable) details before connecting the customer to an agent. Forrester said Hyatt saw a 33% reduction in handle time, as agents were only required for more complicated issues. And, they added, Hyatt expects to achieve higher agent engagement and lower attrition rates as a result. (When playing Centaur Chess, players no longer need to have memorized hundreds of chess openings and detailed variations on their own, so they can spend their time and energy devising more innovative strategies.)
4. Intermingled workflows:  In this model the customer would be routed to a “triage bot,” whose purpose would be to take routine information and assess its complexity before routing the customer to an agent, and/or the customer could be transferred to a bot for routine data capture, following a conversation with an agent. At Hyatt, for instance, an agent can get the gist of a customer’s request and make the appropriate arrangements, but then connect the customer to a chatbot to capture the actual names and addresses of guests, exact dates and check-in times, and credit card data. (Think of a chess computer identifying the most frequent losing moves, given this particular position, based on the whole database of advanced tournament play.)

Ironically, many experts view the computer’s triumph in chess not so much as a threat to human play and ingenuity, but as a boon for it, because when everyone has a computer, chess play will become more “democratized,” because even moderately good players will be able to play with brilliance, unleashing their natural creativity with less fear of making some avoidable error.

And when it comes to teaming AI with frontline, customer-facing workers, we can envision a similar future, allowing workers to use their natural, empathetic humanity to relate to customers more emotionally, with less fear of failing at the routine tasks that also must be accomplished flawlessly.

----

This is just one of the topics I’ll be covering in the 3-day “WOBI on Customer Centricity” virtual event, Sep 29-Oct 1. This is a live, real-time event, so if you have questions, ask me, and I’ll have answers. You won't be consigned to a chatbot! Register here: https://bit.ly/3jI1U5O