Is It Worth Investing in a ChatBot? A Possible Business Strategy

Gist: discussion about replacing human service representatives Recommendations: optimize the app, fill in gaps with a fully trained ChatBot

November 2016: we have a newly elected president, and there is as much suspense about the future of our country as there is about the role that Artificial Intelligence is about to play in our lives. While the presidential election is over, the technology election campaign for the most usable and useful AI gadget is just heating up, and we will all be voting with our dollar.

A possible ChatBot Strategy

From the point of view of business investors, ChatBots have to be profitable to have a chance. One obvious and far-reaching opportunity is the replacement of human service representatives who can cost businesses anywhere from $1 to $30 (or more!) per call/transaction, all things considered.

Even with outsourcing, the costs are substantial, and include calling center maintenance, training of the representatives, insurance, support — all that does not come to mind when all we need is a minute or two of someone’s time. All these costs could be reduced drastically by installing a ChatBot that will require only one large initial investment that will scale through time and through replication, incurring relatively small maintenance expenses. The financial potential is enormous!

However, in order for the ChatBots to fully replace human representatives, they have to be trained and be able to talk and solve problems just as well as the human representatives they will replace. In other words, ChatBots have to be using NLP (Natural Language Processing) and, essentially, pass the Turing test, certifying them to be “as good as” humans for the purpose.

Right now, ChatBots are far from being able to replace human representatives.

Does that mean that it is over for ChatBots? Of course — NOT.

It only clarifies an attainable goal: they have to be trained well enough in human discourse and taught the domain information. Well enough to replace humans.

…and it is just a matter of time.

Importing the research from human representatives.

What is good about this goal is that all the qualitative and quantitative research that has already been done on training human representatives can be fully and directly imported to creating the personalities of the ChatBots. In fact, ChatBots are superior learners to humans. Once a ChatBots learns something — ChatBots does not forget. ChatBots will not make factual mistakes like humans will, and — if trained enough — very few interpretive mistakes. ChatBots will make much better calculations, and perform much faster searched.

Yes, the potential for ChatBots replacing human service representatives is enormous. The key is ChatBot training.

All they need is accurate human guidance. It is all in our human hands now. It is up to the human teachers to prepare useful lessons for ChatBots, and teach them well.

A caveat.

The need for human representatives usually is rooted in the poor service design or an app design. When users make mistakes because things are not clear or because there are gaps or traps in the app itself, then the human representative is called for to unravel the mess. If a business could tighten up their business and their app, so that there are minimal problems and maximum solution options embedded into the app — then the need for a representative, and therefore, the need for a ChatBot to replace it, would drastically diminish. In order to be able to execute such a usable design, there needs to be UX research done to determine all the possible user tasks and all the possible problems they encounter — as much as possible.

Users have goals, not problems.

Applications offer ways to reach those goals, and everything that gets on the way of achieving those goals is a problem. The more the users will encounter problems that are unresolvable within the application, the more they will search for a representative to help them.

Conclusion.

So, the overall business solution for full automation could be a combination of an optimized application (with least problems and most solution for users to reach their goals) and a ChatBot (to fill in those areas that do need special assistance and that could not covered by the app alone).

In those areas where an app alone cannot handle it, a ChatBot would be a great investment, as a replacement of a human service representative.