Chatbots

Chatbots FAQ [An Introduction To Chatbots, AI, ML, And NLP]

Agnieszka Sienkiewicz
Posted on

Calling all potential new chatbot users!

All chatbot enthusiasts and all those of you who ever had a “Oh, chatbots, I’ve heard something about them” moment.   

Attention please.

You are about to witness a solid “wall of text” comprising of nothing more than questions about chatbots, chatbots, and chatbots again.

Don’t worry, there will be answers to those questions and some fancy graphics too.

Here’s a quick preview of main topics this FAQ covers:

Thank you.

Why The Chatbot FAQ?

Nowadays we can use the Internet to search for pretty much anything. The Internet is a vast collection of websites that can give you (almost) all the answers you seek. This is great, no doubt about it. But there is a downside too: having too many websites to browse can be overwhelming and daunting. 

That’s why I decided to collect the most common chatbots-related questions in one place so that you do not need to search countless pages for answers. 

Disclaimer: Chatbot FAQ ≠ FAQ Chatbot

I would like to draw your attention to a significant difference between “Chatbot FAQ” and “FAQ Chatbot”.

The first one, as explained at the beginning, is a general FAQ about chatbots, while the latter one is a bot that conversationally answers FAQ questions

This FAQ chatbot answers common questions related to shipping & tracking 

So let’s begin with some general introductory questions that will help you to enter the world of chatbots.

[Q]: What is a bot?

[A]: A bot is “a computer program that performs automatic repetitive tasks” (Merriam-Webster). Important keywords here are: repetitive and automatic. Humans can perform a lot of tasks but they are limited by time and speed. Bots, on the other hand, can perform the same repetitive tasks – only faster and with no human hand-holding. 

[Q]: What types of bots are there?

[A]: There are many types of bots. Bot type is based on the specific task it was programmed to do and the environment it was designed to operate in.

For instance, there are Internet bots that crawl websites to fetch and analyze data (web crawlers). Bots can also control video game characters to gather resources or fight hostile non-player characters (farm-bots). And of course, there are bots that interact conversationally with human users (chatbots).

Bots are created and controlled by humans so they can be programmed to perform malicious actions and harmful Internet attacks too.  

[Q]: What is a chatbot?

[A]: A chatbot is a bot type that simulates human conversation (a chat). Its main role is to automate the process of interaction with website visitors and social media followers. Think of the chatbot as a robot that answers questions round-the-clock. Chatbots interact with humans via a (live)chat interface or SMS. Live chat presence on the website is indicated by the chat widget.

[Q]: What is a widget?

[A]: A widget is a small application component that enables users to interact with the chat. It’s like a very compact version of the main application which, when clicked, reveals the main chat window. The idea behind is very similar to a system app icon – you click it and the full version of the program is loaded. The widget is extremely useful. First of all, it lets the visitors know that there is a live chat they can use to contact the customer service agent. And secondly, the widget is much smaller than a full chat window so it takes less space on the screen.

chatbot widget

[Q]: Are there different types of chatbots?

[A]: Yes.

There are generally two types of chatbots distinguished:

  • Rule-based chatbots scan for keywords in inputs they receive to recognize the question. 

Rule-based chatbots rely on keywords. Let’s say you type “I’m looking for kids watch” in a live chat. What the chatbot recognizes are “kids watch” and “looking for” phrases. Then, the chatbot checks its predefined answers and returns a matching reply based on those keywords.

  • AI-powered chatbots leverage Machine Learning and Natural Language Processing to deliver more natural conversation

AI-powered chatbots learn as they talk. The more input and feedback they receive the better experience in chatting they build up. The process is similar to human reasoning.

Let’s say you talk with a friend about something and suddenly you hear “Oh no, that’s not what I am talking about. I meant…” – this is a signal for you to realize that you misunderstood that statement. At this point, your mind makes a logical connection:

the statement + initial reply = inaccurate

so next time when you happen to talk about the same topic again, you will know how to reply better. And this is how AI chatbots work. 

[Q]: What is AI? 

[A]: AI stands for “Artificial Intelligence”. AI simulates natural (human) intelligence: it can learn, reason and correct itself… to a certain extent. AI generally falls into two categories: Weak AI and Strong AI

  • Weak AI (or ANI – Narrow Artificial Intelligence) is a system that was trained to perform some specific task. An example of such a task is chatting with humans (remember chatbots?) – this type of AI can do an excellent job in imitating human conversation but would be at a loss if confronted with any other task 
  • Strong AI (or AGI – Artificial General AI) is a system that possesses some general human cognitive abilities. It means that Strong AI is able to solve a problem it was not previously trained for (and it can do it with no human intervention) 

[Q]: What is NLP?

[A]: NLP stands for “Natural Language Processing”. It is a subfield of Artificial Intelligence and an example of Weak AI that is concerned with interactions between natural (human) languages and computers. To simplify, the goal of NLP is to teach computers to understand human language, to recognize speech and to generate natural-sounding speech. 

[Q]: What is Machine Learning?

[A]: ML (Machine Learning) is the ability of a program (like a chatbot) to learn from inputs it receives. Thanks to ML programs can remember previous and current inputs (to build up experience), analyze them and deliver more accurate responses. Machine Learning uses sets of highly sophisticated algorithms (and a lot of statistics!) that make the learning process happen. ML enables computers to learn to converse in a naturally-sounding language (Natural Language Processing) and to “see” images and videos (Computer Vision).

Build Your Own Chatbot FAQ

[Q]: Do I need to know how to code to build a chatbot?

[A]: Not really. All popular chatbots work out-of-the-box. Even users with basic technical know-how are able to deploy their first chatbot in a matter of minutes.

Coding comes useful only when some advanced integration or customization comes into play.

[Q]: How to make a chatbot?

[A]: The process of chatbot building varies depending on the chatbot provider, but most of them follow the same rule: you connect blocks or icons together into a logical flow.

Tidio chatbots are built with nodes that are connected to each other with lines.

Just imagine a mind map: at the center, there is normally a “big bubble” that represents your general idea. Then you connect the big bubble to “smaller bubbles” with nodes. Those smaller bubbles represent possible outcomes or related ideas. Smaller bubbles can be further broken down into some other bubbles-alternatives and which in turn may be connected to the bigger ones. And this process can go on and on.

Eventually, what you get is an interconnected tree (or graph) which becomes your user’s journey (and conversation) map.

This is a simple chatbot template with four possible outcomes (“small bubbles”). The focal point is the question asked to the customer (“big bubble”). Depending on the answer, a chatbot chooses one out of four predefined conversation outcomes (chatbot responses)

[Q]: What is a user journey map?

[A]: It’s similar to a mind map I’ve talked about previously. It is a visualization of a journey your user must take to achieve a certain goal. In the case of a conversation, the goal is to receive a certain piece of information. Key points to consider are:

  • Customers’ motivation: what made your customers come to your website/blog/e-store in the first place? What could they be looking for?

If you know what questions your visitors/ customers could have once they land on your website then your chatbot can handle the situation accordingly.

  • Customers’ channels: where does the interaction take place?

It may take place on your website (or e-store, or social media channel). But your customer could enter your e-store through another page than the home page. Would you like your chatbot to send a specific message if the entry point is, for example, a contact page or pricing list? If so, what would be the flow of such conversation and what would you like to gain from it?

  • Customers’ actions: your customer came with some goal in their mind. What steps could they take to achieve it?

Imagine that you sell shoes, and your customer browses a certain pair, then he switches to another page to look up the size chart, then goes back to the product page. How can your chatbot help your customer to make his shopping experience smoother?

  • Customers’ pain points: are there any challenges they are facing? Perhaps they need to get in touch with you but it’s past your working hours? Or they look for a specific product but are having trouble finding it?

It’s your job to put yourself into your users’ (customers’) shoes and figure out what they might need, how their conversation would flow, what would you and them like to achieve.

You do not need to be a programmer to make excellent chatbots, but you must know your customers well to make chatbots worthwhile.

How Do Chatbots Work FAQ

[Q]: What are chatbot responses?

[A]: They are replies you receive from a chatbot. Chatbots can respond in many ways. For instance, they can reply with an image, a card (e.g. about a product), buttons, a carousel or quick replies

There are also voice-enabled chatbots which, as you guessed it, return audio replies.

[Q]: What are quick replies?

[A]: Quick replies are a group of actionable and automatic responses that temporarily appear as part of the conversation. Quick replies can appear in a form of links, buttons, or specific text values. Once a quick reply is triggered by the user, all other replies disappear.

Quick replies in Tidio live chat. You can see five buttons which represent questions about different topics. Each button triggers a response corresponding to the question.
In this example, “I want to install TIDIO” quick reply was triggered. Chatbot prompted a reply and asked follow-up questions (new set of quick replies) to learn more about the customer. The rest of four initial quick replies disappeared

[Q]: Can chatbots see?

[A]: They do start to recognize visual content better and better. Do you remember what I said about Machine Learning? I mentioned that it drives Natural Language Processing and Computer Vision. Text and audio-enabled chatbot types are powered by NLP, while the power of a visual chatbot type comes from Computer Vision

There are already visual chatbots in use that can carry out a meaningful and natural visual dialogue. To illustrate how visual chatbots work, think of a certain piece of a garment, a dress for instance. The better the chatbots can recognize dress elements, shape, and color, the less you need to type to describe your intent. It also means that you receive more accurate and personalized decisions from chatbots.

Chatbots At Work FAQ

[Q]: What chatbots can talk about?

[A]: Believe it or not but they can be real chatter-boxes! Chatbots are “employed” in different industries so you can find chatbot-specialists that will keep you posted about current weather, some will help you choose grocery, or feed you with the latest news. Some of those little guys end up in e-commerce, customer service, finance, HR, and banking. While others can talk all day long about the meaning of life…

[Q]: How chatbots can help me with customer service?

[A]: Quite a lot. Just think of them as your virtual customer assistants. Chatbots can do all those small things that make a big difference: they can greet your visitors by their name, thank for their placed order, or ask about their wellbeing. They can also place orders for your customers, offer a helping hand with product availability, search for a specific product, update on order status, answer FAQ questions, collect leads, or even save abandoned shopping carts or registration forms! 

[Q]: What is the best chatbot?

[A]: There is no definite answer. Just like not all businesses’ needs are the same, no chatbot is the same. So the best chatbot is the one that can solve your business problems. Ask yourself, what do you expect the chatbot to solve? What is the budget you can allocate to employ a bot?

Does the chatbot provider offer something more than just a chatbot?

We do. What you get from Tidio is a live chat integrated with chatbots. Live chat alone has quite a few features our customers love, and chatbots only add extra possibilities for you to make your service even better. 
And that concludes the following Chatbots 101 FAQ.

If there are any other questions you would like to be answered in the article, do let us know in the comments below!

Agnieszka Sienkiewicz

Agnieszka is a Content Writer at Tidio. She loves running-cycling-swimming, sci-fi movies, and novels. In spare time, she follows the latest news about AI-related stuff.

Similar articles

Chatbots
Chatbot Trends And Stats In 2019
Chatbots
The Best Innovative Chatbot Examples by Industry
Business Growth
What Are Chatbots? 7 Reasons It’ll Make Your Business Grow