Chatbots are not very different from the usual bots that we know about. They comprise of a piece of software which helps in automating a specific task. The only noticeable difference pertains to the fact that these chatbots deliver the task around a single function, namely chat, or simulated conversation. According to Drift, the makers of Driftbot, bots fall under two categories — the programmatic kind with a human-to-chatbot conversation or the intelligent switchboard, required to minimize the interaction.
Most importantly, chatbots generally use machine learning to gather conversational cadences. This enables the chatbot to mimic human conversation, and react to spoken or written prompts to deliver a service. Essentially, chatbot serves as a user interface which reflects the ability to plug into several data sources through APIs. This enables a chatbot to deliver information or services on demand, for say weather forecasts or breaking news. According to Dotan Elharrar, Product Manager @Microsoft AI & Research, bots fall under seven key categories and each category represents how the particular sector would be disrupted.
Classification of Chatbots based on the value proposed
According to Elharrar, we have come to realize that all these chatbots have an associated value. In other words, they are all designed to accomplish a task, thus imparting value to users. Based on this value or nature of job, the chatbots can be broadly classified into 7 key types.
The Optimizer: Most of the other categories of chatbots have spawned from this largest category of functional bots. This chatbot is tasked with concrete challenges, which it solves in a better way than existing apps and websites.Optimizers also makes up the largest category of functional bots, and many other categories are spin-offs of this one. However, majority of these bots fail to keep the promise of delivering better experience than the alternatives. Bot makers should measure and consider the time it took users to get what they came for, while designing or updating the chatbots.
The One-Trick Pony: These include all the mini-utilities associated with a messaging interface that allows users to turn an image into a meme, or piece of text into a video. These chatbots, sometimes reflect excellent cognitive capabilities, and some of them showcase the potential to be disruptive with time. The disruption can be exemplified by Snapchat’s ‘simple’ spectacles.
The Proactive: These chatbots specialize in furnishing the right information at the appropriate time and place. As long as these bots don’t annoy the users with useless notifications, they prove really useful for narrow use-cases. Foursquare’s Marsbot, Weathercat Poncho, and KLM’s bot are some of the most prominent examples in this series.
The Social: This set of chatbots are also designed to accomplish a task. What differentiates these types of chatbots from others, is their ability to build on the power of a group or the crowd. These bots make use of the unique nature of messaging platforms in executing their task.
Some of the most notable examples of this type include Swelly, Sensay, Tinder Stacks, Fam, and Slack bots. Social bots reflect the potential to truly grow viral, even for simple use cases.
The Shield: Essentially a sub-category of the Optimizer chatbots, these help users in avoiding unpleasant experiences. The aim of building such chatbots is to introduce the idea of a friendly chat with these bots.
Some of the common tasks that the bot can execute include fighting a parking ticket, customer support issues, and fending-off obnoxious suitors. These bots also boast about their ability to get things done compared to alternatives, just like in the case of Optimizers.
The Chatty: These chatbots are built to chat with users for the sake of conversation and engagement. One of the most popular chatbots, Xiaoice that comes so close ti being a human has formed a massive user base employing this approach. Automated celebrity and personal bots are some of the examples in this category. Of late, Microsoft debuted its new entertainment chatbot called Ruuh for the Indian market which serves up Bollywood info and more.
Additionally, these bots reflect the capability to truly become opinion leaders. With these, the bot designers can opt to accomplish social or commercial goals throughout the conversation.
The Super Bot – These bots are essentially evolving as platforms that reflect the ability to launch a variety of skills/actions (bots). This set of chatbots will help in addressing the bot discovery challenge, with time. They are also called intelligent personal assistants, and the formidable examples include Alexa, Siri, Cortana, and Allo.
Here are a few groundbreaking chatbots:
Instalocate: A flight tracking chatbot for Facebook Messenger. It tracks flight details using flight number and notifies travelers about flight delays.
Meekan: This chatbot runs on Slack, allowing users to connect with everyone, on any calendar, in a hassle-free manner. Essentially a cross-vendor calendar scheduling platform, Meekan also understands human-readable requests.
Foxsy: The social matchmaking chatbot runs on Facebook Messenger, allowing users to find a meaningful connection with the right person.
BFF Trump: The chatbot was developed to draw attention towards the hateful rhetoric of Donald Trump. The objective was to engage more people to vote their values in the 2016 election.
ChatShopper: This is the first ecommerce fashion and shopping chatbot for Facebook Messenger. It asks users about their fashion preference, and the chatbot replies with the product suggestions.
TechCrunch: The chatbot helps user stay on top of stories and news they care about. Users can subscribe to various topics, authors, or sections of the site, and the chatbot will send you news articles relevant to your subscriptions.
Swelly: It is a social networking chatbot which helps you make decision with your friends. The chatbot recently crossed a milestone with 1.8 million registered users.
Trim: The chatbot helps in analyzing user’s transaction history to reveal details about all the subscriptions. The user can utilize the app to cancel any subscription, for which he doesn’t want to pay anymore.
ButtersBot: The chatbot brings to user the funniest and the hottest polls happening in the world right now. The chatbot likes talking about celebrities, sports, and more. Additionally, it delivers results right to your inbox at the end of each poll.
Letstock: This chatbot has been specifically built for Stock Market Quotes, News and ideas.
Any chatbot conversation often brings up the famed Turing Test, set up by renowned mathematician and computer scientist Alan Turing in 1950 that puts to test the machine’s capability of acing human behaviour. Want to know how humanlike your chatbot can be, test it out at the annual test that also helps in putting AI forward.
Try deep learning using MATLAB