The Bulletin Blog

What Are Chatbots and What Can They Do?

What are chatbots?

“Chatbots” may be a buzzword, but at their core, they are one simple thing: automation systems.

Chatbots are software programs that automate tasks that would otherwise require human labor. Unlike other forms of automation, chatbots are specifically designed to provide automated services in a conversational format.

What can chatbots do?

Chatbots can be designed to meet a variety of business and consumer needs, making their use cases extremely varied.

Different chatbots can help users complete distinct functions such as search for travel deals, set reminders, or find movie recommendations. They can support customer service teams, assist with banking operations, and can even help doctors collect medical information from their patients.

For publishers and journalists, chatbots can act as media distribution tools, connecting users with engaging content and news.

What types of chatbots are there?

There are different types of chatbots with a wide range of sophistication, ranging from bots that reply to simple queries to bots that attempt to offer a high level of automation and understanding via machine learning and artificial intelligence integrations. The type of chatbot that a company might choose to invest in is largely influenced by its intended use case.

Here’s a quick introduction to chatbot classification:

  • Query bots — Query bots are advanced knowledge bases that help users find the answer to simple questions. These bots are best used in customer service scenarios and on websites with live chat software.
  • Intelligent bots — Intelligent bots are bots based on Artificial Intelligence (AI) and Natural Language Processing (NLP). These kinds of bots attempt to understand the nuances of human speech, and to learn from human behavior and responses as they go along. Unfortunately, NLP technologies still have a ways to go; when used alone, their average precision rate is 60-70%, which isn’t very stable for most use cases.
  • Scripted bots — Scripted bots are chatbots that take users through a step-by-step process of engagement. These chatbots use decision trees – or chatbot journey maps – to guide users from their initial interaction with the bot towards resolution. Scripted bots can be applied to many use cases, such as guiding users through content, providing sophisticated customer service, or fulfilling user requests.

Which chatbot is the best?

In the last couple of years, bots based on AI and NLP have been launched with little success, given that users expect bots to carry intelligent conversations. Instead, these types of conversational interfaces have contributed to inflating user expectations and fueling the hype – and in some cases, the subsequent disappointment – surrounding chatbots.

Self-described technologist and CEO of Five9, Rowan Trollope, offers the following insight into the result and origin of this problem: “Developers figured out just how hard it is to make a really good conversational user interface. Product folks were tricked by the trio of Alexa/Siri/Google Assistant into the belief that a conversational interface is easy.”

For an example of the challenges associated with intelligent bots, we can turn to tech giants Microsoft and Facebook. In 2017, Facebook’s own AI bot hit a 70% failure rate because its technology was not sophisticated enough to resemble a genuine conversation, and it failed to correctly respond to the majority of queries.

In an earlier – and more extreme – example, Microsoft launched their AI chatbot, TayTweets, a Twitter bot that the company described as an experiment in “conversational understanding.” According to the bot’s Twitter account, “The more you talk the smarter Tay gets,” learning to engage people through “casual and playful conversation.” Less than 24 hours after its launch, Tay adopted the worst features of its users, spewing racist tweets and rants before Microsoft took the bot offline.

Human conversations are complex – and online, they are often unfiltered. We communicate with emotions, tonal inflections, accents, and contextual information. Even with NLP and AI, the nuanced ways in which humans communicate are immensely complex to imitate, and it might take years of training to perfect. For this reason, it is best for companies to avoid intelligent chatbots while this technology is still in its preliminary stages. Instead, we recommend using scripted chatbots for content delivery.

Scripted chatbots keep user expectations in check by guiding readers along a predetermined set of paths (sort of like a “Choose Your Own Adventure” game). They are easier and cheaper for developers to create, and their abilities depend on the quality of their content rather than the current stage of their technology.

Where can I find chatbots?

Chatbots can interact with users on a number of platforms, such as through an app, on a website, on collaboration workspaces (such as Slack), or via messaging programs.

Depending on a chatbot’s purpose, it may perform better on a certain platform. For example, query bots that are designed to assist online shoppers with their concerns (or connect users to customer service agents) often fulfill their purpose most effectively on a company’s website or mobile app. On the other hand, if an organization’s goal is to reach a wide audience and stimulate user engagement, then scripted chatbots on messaging apps are more effective.

Many messaging platforms have actively encouraged and facilitated the creation of chatbots by publishers and other businesses. Telegram Messenger, for example, has become known for their easy-to-use system for chatbot creation and registration. This is thanks in part to the BotFather, a chatbot designed by Telegram to guide users through the bot building process. As Telegram explains: “BotFather is the one bot to rule them all. It will help you create new bots and change settings for existing ones.”

The Botfather. Telegram. May 2018.

TFacebook’s messaging app, Facebook Messenger, is another messaging platform that has actively endorsed bot building. As of May 2018, the company reported that there were 300,000 active bots engaging with customers on their platform – three times the 100,000 bots that were active a year ago. According to Facebook’s Vice President of Messaging Products, David Marcus, these chatbots facilitate the exchange of over 8 billion messages between customers and businesses on Facebook Messenger each month.

Who has chatbots?

The announcement that messaging apps were adopting chatbots was a green light for media companies. Publishers like CNN and Business Insider were ready with chatbot launches directly following Facebook’s bot announcement at the F8 developer conference in 2016.

Since then, media outlets of all sizes have worked their way towards becoming a part of this platform evolution. Quartz, a business news startup, recently announced the release of the Quartz Facebook Messenger chatbot at their 2018 SXSW event, despite having their own in-app chatbot since 2016.

There is no limit to who can adopt a chatbot. Other media organizations that have adopted chatbots include the BBC, Forbes, National Geographic, and more.  

How do you use chatbots?

Not every media company that has adopted chatbots has unlocked their full potential. Companies that take on generic chatbot software may have a hard time adapting chatbots to their particular use cases. Others may have jumped the gun on chatbots, without being aware of how exactly to put them to their best use.

Ultimately, the way that chatbots work best for publishers is by using them as a sort of interactive newsletter – think Mailchimp for messaging platforms. With the right chatbot software, companies can create bots that send their audience guided stories using photos, text, emojis and more. These stories can be used to boost engagement, generate website traffic, and more.

So, why chatbots?

Chatbots offer a way to deliver automated content directly to audiences. Messaging bots (chatbots that are available over messaging apps) aren’t hindered by third party forces like social media algorithms – which could cut reach. They are sharable and even monetizable. Chatbots put control of the narrative back into the hands of publishers. They give them access to metrics and insight into what content their readers are interested in. In the end, chatbots are the perfect tool for delivering what media companies are looking for – engaging content packaged for today’s audience.

For more information about how chatbots can help your media organization, check out our blog post, 5 Important Reasons Why Publishers Need Chatbots on Messaging Apps

This blog post can be found in our eBook, Streamlining Content Delivery with Messaging Bots. For a free download of this eBook, click here.

Xavier Orduña