The Five Types of Chatbots.

T

George Henry Lewis once said, “Science is the systematic classification of experience.”

Only things that can be classified and codified are allowed to enter our realm of understanding and acceptance. And given the divide between whether 2018 will be the next reckoning or the death of chatbots as an entity, we take a much more moderate stance; Chatbots are now a part of our lives.

So in order to expedite this process, welcome to the five broad (and possibly slightly inaccurate) classifications of the Bots that most people deal with on a daily basis.

1. Simple Use Case bots

One of the most common occurrences you’ll encounter on platforms like Telegram or Slack. These bots are built with an easily attainable end goal in mind, like to tell you the weather.

A personal favorite of the Verloop team is ImageBot, a bot that fetches images and gifs for the user. So if you feed “happy dog” into the prompt, voila! Smiling dogs fill your chat feed. Donut is another cool bot on Slack. Donut encourages employees who don’t know each other to grab a coffee through a randomized roulette program.

Don’t be fooled by their face-value simplicity though, some of these bots are an impressive display of the capabilities of code in the right hands. Project Murphy springs immediately to mind, originally prototyped by a team at Microsoft at their Build Conference to show off Azure’s Machine Learning abilities. That little experiment went viral, with over 90 million people uploading images of themselves to have the bot guess their age.

Project Murphy is the 2.0 version of version ‘HowOld’ robot and can answer questions like “What if Gandhi had a beard?”, “What if Beethoven was a rock star?” and “What if I looked like Mona Lisa?”

2. Ease of Use bots

As with all technology, from Uber to Airbnb, the intrinsic purpose of advancement is to make life easier. Uber made the process of getting a cab a lot more hassle-free, Airbnb helps users find cheaper accommodation with ease and Spotify revolutionized the music market with streaming.

In a world where most UI’s rank experience at the cost of ease, chatbot present companies with an intriguing opportunity. Repetitive tasks, like FAQ’s, customer service or online shopping that takes an existing UI more than two steps, is where chatbots thrive.

Bots designed to reduce the number of steps a specific process takes, and that focus on an “ease of experience” come under this umbrella. A great example comes from this article titled Chatbots for eCommerce — The Who, What, Where, When and Why

A simple request like “pink running shoes” can be executed in a matter of seconds, as opposed to the tedious process of filling out filters yourself. *

*After typing that out, I actually tried it out myself on Amazon’s Indian mobile app. It took me 2 minutes and 7 seconds, try it out.

As opposed to a simple scrolling mechanism, where the user has to go through hundreds of unnecessary results, chatbot based support can take a user directly to what they’re searching for, without the static noise of pointless substitutes.

3. Troubleshooting bots

One of our most favorite people, not simply limited to the tech world, is Joshua Browder. You may have never heard of him, but if you’re interested in chatbots, I highly doubt you’ve never heard of his contributions to conversational automation. See, Joshua is the founder of the now infamous DoNotPay bot, code that helped give the legal world a much-needed defibrillation.

In an interview he gave to the Independent, this at-the-time 18-year-old, had started driving and began to incur numerous parking tickets. In what would be the buildup to the first large-scale Troubleshooting Bot, he formed the perception that parking tickets were disproportionately targeting the elderly and disabled. After noticing the “formulaic nature” of the process by which they could be appealed, Browder created a chatbot named ‘DoNotPay’. Since its launch, the site has attracted over 175,000 successful users. It has saved UK and New York motorists an estimated $5 million.

And much to, what I imagine is the dismay of thousands of soon-to-be law graduates, he didn’t stop there. He went all in, doubled down and beat the house.

In the spirit of troubleshooting, Browder pushed out another 1,000 legal bots, across the 50 U.S. states and the U.K. to help people fill out legal documents. He thereafter went on to make a bot to help people sue Equifax for their massive cock-up and yet another bot to help refugees seek asylum in US and Canada, with asylum support in the UK.

But my absolute favorite part of this story comes in the form of a quote from the aforementioned Independent article.

With all the growth, it would seem natural for DoNotPay to take on venture financing and contemplate monetization, but so far Browder has pushed off VC financing and has been insistent that his creation will remain free.

Let’s be honest, we’ve all needed one of these bad boys in our lives. And they’ve been a string of bots that fall under this category since. MSSG Voting Bot helped people find their nearest voting location easily. Ghostbot helps you shoo suitors who just can’t take a hint. Bloodlink is an Indo-Danish that helps assist people with blood donations.

4. Open Ended bots

In an article called NLP, NLU, NLG and How Chatbots Work, I make the case that with the improvement in Natural Language Understanding, we’re on the brink of some incredibly personable and human bots.

I was wrong.

We’re already there.

There exist many Chatbots on the open market that just want to talk, bots that more than anything want to beat the Turing Test. But there are some chatbots that do so much more.

Woebot is the first example of change. Created by scientists and tested by means of a randomized controlled trial at Stanford University, Woebot is an automated conversational agent (chatbot) who helps you monitor mood and learn about yourself. In simple terms, a therapist. Which works.

In the study, using Woebot led to significant reductions in anxiety and depression among people aged 18–28 years old, compared to an information-only control group. 85% of participants used Woebot on a daily or almost daily basis.

You can read their entire peer-reviewed study here.

To understand why this huge, it is important to note the following. Therapy is hard. Therapists are rigorously trained to be human antennas, their work is highly interactive, based on thousands of hours of supervised practice in order to become licensed to practice on their own.

Some are incredibly specific. An app called Karim counsels Syrian refugee children; Emma helps Dutch speakers with mild anxiety, and MindBloom allows users to support and motivate each other.

And this isn’t the only hoop Chatbots are finding their way through.

If you Google “smartest chatbot”, you unsurprisingly get Mitsuku. Mitsuku is an award-winning chatbot created by Steve Worswick. It has won the Loebner Prize, which is awarded to the most “human-like” Chatbot, three times (in 2013, 2016, and 2017).

Mitsuku is created from AIML (Artificial Intelligence Markup Language), an XML dialect developed by Richard Wallace and a worldwide free software community between 1995 and 2002. AIML formed the basis for what was initially a highly extended Eliza called “A.L.I.C.E..” (“Artificial Linguistic Internet Computer Entity”), another startlingly open-ended bot, which won the annual Loebner Prize Competition in Artificial Intelligence three times and was also the Chatterbox Challenge Champion in 2004.

All of these achievements are powered by the incredible Machine Learning capabilities that bots possess, which sometimes, are used against them. The most popular name is this list, is, of course, the rise and downfall of Tay.

5. Virtual Assistant bots

You know what these are, and have undoubtedly interacted with them at some point of time. Meet your Siri’s, Cortana’s, Google Assistants and Alexa’s.

The most widely used in the VA’s on mobile platforms in the US were Apple’s Siri (34%), Google Assistant (19%), Amazon Alexa (6%), and Microsoft Cortana (4%).

Apple and Google have large installed bases of users on smartphones.

Microsoft has a large installed base of Windows-based personal computers, smartphones and smart speakers.

Alexa has a large install base for smart speakers.

Virtual Assistants have been well documented; they’re hyper-smart, always at your service and ripe for monetization. Leading the show though is Amazon.

Amazon sales increased by 38% in the last three months of 2017, to more than 60.5 billion dollars, as the technology giant reported its latest quarterly financial results. Amazon founder and chief executive Jeff Bezos said the rise of the company’s Alexa products was a key factor in the company’s continued growth.

“Our 2017 projections for Alexa were very optimistic, and we far exceeded them. We don’t see positive surprises of this magnitude very often — expect us to double down,” he said. “We’ve reached an important point where other companies and developers are accelerating adoption of Alexa. “There are now over 30,000 skills from outside developers, customers can control more than 4,000 smart home devices from 1,200 unique brands with Alexa, and we’re seeing strong response to our new far-field voice kit for manufacturers.

According to Amazon, the Alexa-enabled Echo Dot, along with the company’s Fire TV Stick streaming device were, in fact, the best selling products across all of Amazon in 2017.

About the author

Anush Fernandes

Loves Canines, Conversational automation, and Curry. Stephen and otherwise.