Chatbot mania: double or bust?

Image of Star Wars robot

By Tim Wouters, senior UX Designer at Internet Architects

You’ve seen them around the block. Chatbots were plastered all over 2016 like Donald Trump’s face over Blue America’s nightmares. Will their reign as the new UX paradigm continue into 2017?


There’s been no lack of enthusiasm when it comes to experimenting with chatbot technology. Retailers, banks and news outlets seem convinced that digital assistants are the future of personalization. Tech giants such as Facebook, Amazon, Microsoft, IBM, Google, Apple, Slack, Kik and Telegram are competing in a new gold rush to become the ultimate chatbot platform. Yet, to date, none of them seem to have cracked some of the most fundamental problems that plague our new chatty robot friends. To wit:


1. Frustrating

Admit it: chatbots are frustrating to talk to. Like talking-to-a-2-year-old frustrating. Sure, you can have some basic interactions as long as you choose your vocabulary wisely. But anything beyond casual toddler speech will give you nothing but a dazed expression of non-comprehension.

“Agreed,” you might say, “but that’s only a temporary problem until technology catches up. Right? I mean, just look at that all-star team of big tech companies trying to crack the problem.” Well, ok. I guess throwing some serious AI (Artificial Intelligence), big data, deep learning, neural networks, sequential inference and what have you at it might give us some progress. But then again: tell me how today’s personal assistants are any better than what IBM’s Eliza demonstrated 50 years ago? Given how microprocessors have progressed since then, have chatbots kept pace?


Eliza Chatterbot - A Program that can Chat with Humans

Have chatbots really evolved since IBM’s Eliza? (source: PlanetSourceCode)

2. Awkward

If Google Glass has taught us anything, it’s that there’s a big gap between “nerdy-kinda-cool” and socially acceptable. I mean, who wants to be caught in public talking to a robot? We don’t treat machines the same way we treat humans. Machines are given instructions, not sentences. That’s an important distinction to make, although the difference might blur on occasion.

So why are we trying so hard to make our bots sound as human as possible, thereby violating every design principle of Masahiro Morti’s Uncanny Valley theory? Machines don’t have emotions, so we don’t have to cater to those emotions when interacting with them. A couple of keywords will do, just enough to make the thing understand what you want. There’s a good reason why Google’s list of top searches in 2016 still looks like a Wikipedia summary page. Because that’s all ‘Googlebot’ needs.

Screenshot Google search without results

Even Google remains conspicuously silent when talked to in a human fashion

3. Complex

Many of the processes supported by chatbots today try to replace what used to be known as a ‘user interface’ and cram it into a much more limited ‘conversational UI’. And that causes confusion for the end user. Familiar interaction signposts – known as ‘affordances’ – such as navigation menus, search boxes and buttons are gone. Instead, there’s just a single blank text input field staring at you, waiting for your commands. Where to start?

I’m afraid chatbots, in the current state of affairs, find themselves on the wrong end of the simplification curve. Removing elements from a user interface only yields advantages as long as it reduces physical effort for the end user. Beyond a certain point, however, those advantages dry up, and any further UI simplification will only increase mental effort for the end user and, thus, complexity. This phenomenon is also known as Tesler’s Law and has been documented extensively by UX guru Jakob Nielsen.

Graph showing The interaction cost of talking to chatbots

The interaction cost of talking to chatbots today (source: Internet Architects)

So, what’s going on?

What is it that makes chatbots so wildly popular despite these inconveniences and obvious frustrations? What’s going on? In fact, there’s a lot going on right now, and chatbots are absorbing it all at once:

1. Conversational UI as an interaction pattern

Whether you call them ‘command line’, ‘chat’ or ‘conversational’, these types of text-based interfaces have been around for decades, but somehow exploded over the past 12 months into a panacea. From simple web forms to news apps, fitness trackers, e-shops and recipe catalogs: no-GUI seems to be the new GUI, to such a degree that we might ask ourselves if we’re not climbing our way up the steep side of the technology hype cycle, and facing a steep descent of disillusion. Only time will tell. Heck, even one of our own high-profile projects of 2016 involved a conversational app.

graph Gartner’s technology hype cycle

Gartner’s technology hype cycle: are chatbots overhyped?

2. Messaging apps as the new platforms

In 2008, the World Wide Web was already a cluttered mess with fierce competition over user attention. Standing out from the crowd was hard and increasingly at the mercy of Google’s search algorithms. Then, Steve Jobs’ announcement of the App Store triggered a whole new swarm of brands trying to get into the user’s pocket and onto the smartphone home screen.

Alas, the app economy well seems to be drying up too. App fatigue is now officially a thing, and smartphone users are massively gravitating towards just a few apps of choice, most of which are social and/or messaging apps. It should come as no surprise that brands vying for user attention are excited at the prospect of these messaging apps opening up their platforms for third-party integrations. Facebook is testing the waters with its messaging app, Slack is well-known for its many available integrations, and WeChat has become a veritable chat-commerce juggernaut in Asia. Even Apple has opened up Siri through a limited number of APIs.

‘Follow the user’ has been the ground rule throughout the Internet era. Today’s no different.

3. Voice recognition

As much as the digital revolution has seeped into our daily lives, we’re only just at the beginning. The miniaturization of general purpose computers-with-screens has reached its limits with the smartphone. Even Apple’s smartwatch is now being repurposed as a dedicated fitness device.

But there are so many opportunities our there to make objects connected and smart, even if they don’t offer the physical space for a user interface as we know it. Thermostats, light bulbs, store shelves, vehicles, roads. The only limit is our imagination. As a precursor to this new Internet of Things, voice input and natural language processing are hot topics right now, with Alexa, Siri, Google Assistant and Cortana leading the way as our faceless chatbot assistants, trying to figure out what we mean and knitting together services to do what we want.

4. Automation through Artificial Intelligence

Compared to the above, the gradual progression of ‘automation’ seems to move mostly under the radar. Going back to printing presses and the industrial revolution, mankind has always sought to make life easier through automation. The advent of the Internet, together with hugely powerful computer networks, have given a new dimension to that process, simplifying our daily routines at a steady pace:


Sure, artificial intelligence and automation will change our society in profound new ways in the decades to come. It will cost us ‘old’ jobs and create ‘new’ jobs. It will allow us to become more productive, work less, and produce the wealth needed to fill that new-found free time.

But most of that automation bliss will happen silently. Gradually. Under the hood. By standalone apps being converted into embedded apps, enhancing real conversations between real people. By humble machines in data centers that get the dirty job done. Not by annoying chatterbots you want to punch in the face. In other words: if any bot is going to rule the world, I’ll put my money on R2-D2, not C-3PO.

2017 might not yet be the year of demise, though.


More news

How to get on top of your website’s performance

Picture of running contest with athlete in starting blocks

A faster website only bears fruit. Like we’ve told before: your customers’ time is your money. As your website loads faster, less people will abandon your website, they will stay longer and ultimately your bounce rate will benefit. A good performing website even impacts SEO, making sure it’ll be easier for (potential) customers to find you. A faster website improves conversion, it increases visitor satisfaction and it reduces visitor abandonment.

more about this news article