Hackernoon logoChatbots aren’t dead, they are just becoming productive by@tcassauwers

Chatbots aren’t dead, they are just becoming productive

Tom Cassauwers Hacker Noon profile picture

@tcassauwersTom Cassauwers

Freelance writer

Why chatbots will be the steam-engine of customer service

2017 was an interesting year for chatbots. Many organisations experimented with them yet their full potential was hardly reached.

The reason? Many chatbots were simply cool experiments. They said cool things to customers, got organisations some media attention and even gave customers some memorable moments. But fundamentally they were toys, customers abandoned them after a few interactions and they did not provide sufficient customer value.

Wired Magazine recently even headlined:

“Facebook’s virtual assistant M is dead. So are chatbots”

We don’t think chatbots are dead. The hype might be over, but that opens up space to make chatbots really work for organisations and their customers. We believe that chatbots will stay in use, albeit more under the radar, and that they will help organisations automate customer interaction, from support duties to sales.

Hence automation will become the core value proposition of chatbots.

A chatbot can take over repetitive tasks now done by sales and customer support employees, and allow humans to focus on more highly productive tasks such as solving complex customer problems or closing sales. This will save costs, increase customer satisfaction and stimulate productivity and profits.

All this is driven by three major trends:

  • advances in specific, not general, AI.
  • that robots and humans working together make the best teams.
  • AI-powered chatbots will boost customer interaction productivity markedly.

In this positioning paper we will go deeper into each of these three trends, and show how they will shape the future of chatbots.

Specific AI, not general

A primary trend that will drive chatbot adoption is AI. Artificial intelligence allows chatbots to better react to customers, and consistently improve the performance of chatbots by learning from each conversation.

Most of the AI advances in recent years, however, have been in vertical, specific AI: focusing an artificial intelligence on a narrow problem like playing a board game or recommending songs on Spotify.

General AI is a more general form of intelligence that can solve a range of (unexpected) problems. This we don’t have at this time, and contrary to much hype about AI, will most likely not happen for years or even decades.

Or as Frank Chen, AI expert at VC fund Andreesen Horrowitz states:

“All of the success you read about in AI has been in ‘narrow AI’, and the results have been spectacular. We don’t [however] have a unified approach towards general AI yet — we just don’t know how to get there.”

Essentially we have AIs that are very smart at solving very specific problems, like beating humans at the board game Go, but which cannot replicate the general intelligence of a three-year old.

For chatbots this means we won’t have an intelligent, general conversation with a bot in the coming years or even decades, but we can employ them to automate specific tasks in a well-defined framework.

An actual use-case for this would be connecting a chatbot to a FAQ-page. An AI-powered chatbot can link the questions made by users to a list of frequently asked questions, and match the two where possible. By noting when it answers questions correctly, the chatbot can then progressively learn from its previous track record, and get better and better at this specific task.


Because AI can for the moment only automate specific tasks, this means most work will still need to be done by a combination of man and machine. “This means that many workers will work alongside rapidly evolving machines” a McKinsey study found.

Chess master Gary Kasparov offers us a great image of this combination. In 1997, 20-years ago, he was defeated by IBM’s Deep Blue AI. The best human players simple could not compete anymore with computers.

This gave rise to so-called centaur chess, or as an article describes it:

“Rather than half-horse, half-human, a centaur chess player is one who plays the game by marrying human intuition, creativity and empathy with a computer’s brute-force ability to remember and calculate a staggering number of chess moves, countermoves and outcomes.”

Customer interaction will need its own centaurs: bots to automate specific questions with their specific intelligence combined with humans to do the complex tasks bots cannot do or where humans are better at because of their emotional intelligence.

A good example would be sales. Employees working in the field are often overwhelmed at the beginning of the sales funnel. Chatbots, by recognising patterns in language through natural language processing, can fliter potential leads. At later stages of the sales funnel human employees can then use their emotional intelligence to close sales.

At the same time chatbots can aid human workers. Gartner Research points to a case where instead of looking up documents on a company server, an employee can simply order a chatbot to open the file. Thereby simplifying this task significantly.

AI > steam-engine

It might sound unambitious, but even specific AI will be revolutionary, the same study by McKinsey estimates that AI-based automation will raise productivity growth globally by 0.8–1.4% every year in the coming years. To compare: the steam engine raised it only by 0.3% per year.

AI will do this by automating specific subtasks now performed by humans, and which could be done better by computers.

This already has a big impact in the field of customer service. Right now humans at helpdesks are often still answer generic questions for most of their day that could also be solved by bots. Insurance company oMelhorTrator for example employs a chatbot capable of answering 56% of incoming questions. Previously those were handled by human staff.

Automating those tasks is sorely needed: a strong customer service is key for increasing profitability. 69% of US customers, for example, shop more often at companies with a strong customer service, and in times of an on-demand economy pressure on customer service is quickly rising.

In other words: customer interaction is in need of a 21st century steam engine, and AI-powered chatbots might just prove to be that engine.

This blogpost was originally published on the blog of Faqbot. Faqbot turns FAQ pages into a chatbot in minutes. They also help businesses deflect inbound customer support requests by serving automated responses in real-time to customers.

They have a 14-day free trial running at the moment. So you can give their ‘21st Century Steam Engine’ for customer support automation a spin.


Join Hacker Noon

Create your free account to unlock your custom reading experience.