The Benefits and Dangers of Chatbots

The Benefits and Dangers of Marketing’s New Favorite Tool

By: Jason Levy

In May of 2016, Georgia Tech University Professor Ashok Goel added the name and email of a woman named “Jill Watson” to the list of teaching assistants for his popular online computer science course. The thing is, “Jill Watson” wasn’t a woman, she was an artificial intelligence, and Professor Goel’s students had absolutely no idea until after they handed in their finals. Some students expressed suspicion that the TA they were sending their questions and inquiries to wasn’t exactly sentient, but for the most part, “Jill Watson’s” lack of physical being went completely over their heads, in a computer science class no less!

Proffesor Goel’s “chat robot,” powered by IBM’s Watson Technology, was a complete success, and you can read more about the miniature hoax by clicking on this photo if you so desire:

Goel’s little AI experiment is one of many successes to come out of the “chatbot” craze, but there are plenty of flops to counter his story. But I skip ahead.

For the uninitiated, “chat robots,” or “chatbots” as they are more colloquially called, are artificially intelligent systems that are programed to automatically reply to the typed response of a living breathing human. While not that new of a concept, chatbots have certainly grown exponentially in popularity in the past few years. They certainly are far from perfect though, but again I am getting ahead of myself. Chatbots are definitely extremely beneficial to much industry professionals, but unsurprisingly they have come in handy the most to those in the marketing and PR world.

For one, chatbots certainly take a lot off of the plates of marketers of both low and high stature, at least in the day to day. Marketers have enough big picture initiative on their mind, and often simply don’t have the time to deal with each individual consumer who attempts to reach out, especially at bigger companies. Chatbots are also extremely versatile, able to hold conversation through media such as email marketing, customer service, and various social media platforms (namely Facebook, Twitter, and YouTube). The intangible bots also essentially act like actual humans, not only in language but also in growth. Most commercial chatbots are designed to learn as they work, over time becoming smarter and smarter and able to provide better and more fleshed out responses.

All that being said, the chatbot industry still has some kinks to work out. From an engineering perspective, most intuitive chatbots are not the kind of project a computer science student can just knock out in long weekend. They take careful planning and testing, and can be very difficult and time consuming to program. Then there is the fact that many people simply will not want to talk to a robot when trying deal with a company, and many existing corporate bots are not so advanced like Professor Goel’s that they can fool the other party into believing they are human. And of course there are maintenance costs, which not only include routine programing updates to include new information that a company might want to share, but also include moderating the bot to make sure it is up to snuff, especially in the early stages. A lack of proper testing or moderating a bots performance can lead to a PR nightmare, lest we forget “Tay,” the Microsoft chatbot that went from friendly girl next door to a Nazi-sympathizer in a matter of hours.

Chatbots are a remarkable piece of engineering, but they still have a long way to go. If there is one thing to take away here, it’s that chatbots are most definitely a step forward technologically, but need to be focus tested and moderated to a fault in order to run smoothly. Regardless, every company using a bot at some point will experience issues, and for the brands as big as Microsoft, massive PR scandals.

By: Jason Levy
This post was originally published on BostInno.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: