When the hype around chatbots started, many myths appeared around them. As the technology developed and was promoted some myths vanished, some stayed, and new misconceptions are still emerging.
We gathered the most popular of them and explored what they are based upon and how much truth they contain.
1. Chatbots are a new technology
In fact, the history of chatbots has its roots in the distant past.
The first member of this “family” was built in 1966 – it was ELIZA. This computer program created by Joseph Weizenbaum is known mostly for the script simulating (or rather parodying) a psychotherapist. It used users’ input to formulate non-directional questions and hold a conversation. Afterwards, there were PARRY(1972), Jabberwacky (1988), Dr. Sbaitso (1992), A.L.I.C.E. (1995), SmarterChild (2001), and at last, IBM Watson (2006), Siri (2010), Alexa and Cortana (2014-2015). In 2015, Telegram launched Bot API for developers, Slack launched 150 bots and invested $80 million in the development of Slack chatbots. In 2016, Facebook launched a platform for Messenger bots. So, this technology is relevant and developing but not new.
2. Chatbots always pretend to be humans
This myth has some basis in fact. On one hand, it might be caused by a famous Turing test. A machine is considered to pass it if tricks the judge into doubts or mistake about which of two (hidden) responders is human.
On the other hand, there was a time when chatbots representing business introduced themselves as live operators. Now such virtual agents are considered bad manners and become a thing of the past. Today, one of the principles of chatbot UX design is not to try cheating on users in that way. Modern bots for business are more like apps and don’t try and pretend humans.
Moreover, it is difficult to do technically. Not many bots can plausibly simulate a human, especially if they have to complete certain business tasks and are not created just to pass the test. At least, you can’t avoid phrases like “I’m afraid I don’t understand you”, “Could you repeat it in other words” and “I’m only learning to talk to humans.”
3. Chatbots interact through the “conversational” interface
The first bots were indeed more conversational. But now most of them can send images, video and even have control elements (buttons and menu).
There are bots whose only job is to find gif’s, create stickers, download music and convert files. You needn’t conversational interface to do it. There are even flow-oriented chatbots that interact with users with the help of buttons only and user input is disabled. You cannot even type a message to them.
Graphic elements are familiar and convenient for the majority of people. Also, flow-oriented bots with buttons who don’t try to “have conversations” with users and don’t claim to understand natural language, often make fewer mistakes about what the user wants and complete similar tasks quicker than their “conversational” brothers.
4. Chatbots will kill apps
Well, chatbots make businesses accessible on mobile devices and may have some advantages over apps. A chatbot of your company will display your products and services in a messaging app and allow customers making orders, finding shops and receiving 24/7 tech support. It doesn’t need to be downloaded and installed by users and for a business owner, it can save time and money for development and deployment (speaking of a simple chatbot). But it is improperly to tell that chatbots can replace apps. They have different functionality and accomplish different tasks. It looks like chatbots will complement apps.
5. Chatbots will replace humans (they’ll take away our jobs!)
As a matter of fact, chatbots reduce workload and free human employees. They are good at repeated, routine tasks that annoy and tire people: answering FAQ, receiving requests, lead generation. Due to this, the employees can focus on non-typical and creative work. Bots often perform additional, subsidiary tasks helping staff members, for example, customer care operators, to provide better and quicker service.
6. It is easy (and cheap) to create and launch a chatbot
In reality, it is not difficult to create only a simple bot. Indeed, you can, in a couple of hours, develop a chatbot that will provide contact information and answers to frequently asked questions. Even if you don’t know any programming languages, you can use development platforms to make a bot without coding. Many platforms provide free service though they limit bot abilities. But if you
- need a chatbot for business
- plan certain functionality,
- want admin panel to manage it,
- want it to understand free user input and learn,
- prefer it to work in several messaging apps,
it will be not easy to develop such virtual agent. Such project demands experience, time and money. See more about how much time and money it can take in our article.
7. All the chatbots are artificial intelligence-driven
First of all, AI itself is a source of many myths and discussions (or rather holy wars). Obviously, it is a fertile ground for myths about supersmart chatbots.
There are plenty of flow-oriented chatbots with buttons we have mentioned before. They are not intelligent in principle – they just display predefined answers when a corresponding button is pressed.
There are more complex bots which can “understand” free user input and process natural language but they just extract keywords from phrases. At last, chatbots capable of machine learning exist. They can recognize the requests typed by users considering context (with a probability that can differ but is always less than 100%). An answer can be selected from a database or generated after the machine is trained on a large training set. At best, chatbots use so-called Narrow AI – an intelligence that allows accomplishing certain specialized tasks (so does every system existing for now).
Even the creators of IBM Watson supercomputer, that won Jeopardy!, don’t use the term Artificial Intelligence but call it augmented intelligence to emphasize the aspect of the partnership between humans and machines.
So, there is some truth in the myths about chatbots. But not all of them are very smart, they are not going to replace humans and can communicate not just in conversational form.
Do you want to add some myths to our collection and read an explanation about them? Send your ideas to our email.