When Bots Fail

The fall of Bots as everyone’s favourite toy has been as spectacular as their rise (maybe a bit more). With Artificial Intelligence increasingly occupying mind share, it was natural to use the tools to build these robots that are used for conversation aka chatbots. Unfortunately, conversations are complex as human beings don’t converse in simple and logical statements.

First, there is sarcasm, a trait that is complex to many humans, leave alone bots. Using Chatbots in a domain where there is a lot of complaints is a dangerous place. All the furore in the leading Airline after the below twitter complaint was something bound to happen.

Next is the amount of resources companies use to train the bot as well as integrate it to other pieces of information. A leading bank implemented this chatbot on their website and even after months of being in production cannot answer the most basic of questions.

So What is the Solution?

Chatbots have yet to reach a stage where they can replace humans, so do set expectations appropriately by Introducing yourself as a robot.

Understand the scope of what can be done and make efforts to do it well. Introduce buttons to simplify answers and questions.

Have decision trees and well defined fallback answers.

But most importantly keep improving the bot as it should become better with time.

And finally, if a chatbot is unable to make sense, do handover to a live agent else you would leave the customer more frustrated.

The move to Artificial Intelligence and Bots is almost inevitable with the rate at which investments are ongoing to improve it from a technology perspective. However, understanding humans and customers especially will be at the heart of making the AI successful in real world conversations.

2020-04-02T08:04:11+05:30December 10th, 2017|

Leave A Comment