Conversational AI is Eating the Web

Could the web, as we know it, disappear?

Is it possible that beautifully frustrating browsers and websites disappear? Could these essential tools of daily life and work become technology footnotes alongside Usenet, Gopher, IRC, DMoz as antiquated interfaces to all the Internet’s amazing treasures?

This future may seem hard to imagine considering most of us grew up experiencing the Internet by browsing websites inside of web browsers. But, believe it or not, the browser (1993 — Mosiac) and hyperlinked websites (1989 — HTTP) are relatively new ways to get information from the Internet. So, it’s not inconceivable that we’re about to transition into a new user interface.

“closeup photo of eyeglasses” by Kevin Ku on Unsplash

If you’re curious about what is to come, you’ll have to monitor a variety of discussions. Real innovation in user interface and experience (UI/UX) design is embedded in several related technology trends — Zero UI, Conversational AI, AI Assistants, Conversational Agents (Chatbots).

I like the term Conversational AI because it best describes the underlying technologies that will ultimately enable this new UI/UX. To understand my prediction for Zero UI in the future, let’s better define this Conversational AI domain.

What is Conversational AI?

Conversational AI is the use of messaging, voice, and assistants to communicate with computers in a way that creates personal user experiences. In simpler terms, it’s talking to computers in a natural, human-like, conversational style.

Of course, the challenge is computers aren’t naturally conversational. We have to teach them, which is where artificial intelligence (AI) comes into play. To make a computer conversational, we have to solve three technical challenges:

  1. Natural Language Understanding (NLU) — Computers need to understand the user’s request (intent).
  2. Natural Language Processing (NLP) — Computers need to process that intent into a task to retrieve the appropriate response.
  3. Natural Language Generation (NLG) — Computers need to translate the response into something that makes sense to the user.

Each of these steps requires enormous amounts of analysis and learning (machine learning) fueled by a massive number of diverse conversations (training data set). This collection, analysis, and learning have only recently become realistic at the scale necessary to even get close to making computers conversational.

The ubiquity of mobile phones and the swift proliferation of devices like Amazon Echo and Google Home have made crowdsourcing these conversations, at scale, relatively inexpensive and straightforward. Combine this explosion in training data with nearly free data storage, brought on by cloud computing, and rapid advances in machine and deep learning and you have a perfect storm. Conversational AI, as a superior interface, is becoming a reality.

What is the web?

It turns out that the web is an inferior version of Conversational AI. Think about our current version of the web. To give you reasonably intuitive access to information on the Internet, we force designers, developers, and users (people) to do all of the natural language processing.

We build websites, informed by what we think are the most common use cases — workflows that users will use to get to what they want from our sites.

(Secret Insider Sidenote: Ironically, that’s not really how we design websites. Truthfully, we design them to “funnel” you to where we want you to go. All along the way coaxing you to do what we want you to do.)

Here in lies the everyday frustrations of using a web browser and navigating websites. This current UI/UX paradigm requires users to find what they want by interpreting what the designer thought when they designed the site. Not ideal.

Why is Conversational AI is a Better UI/UX?

One persistent theme in technology innovation is that users, especially power users, tend to fill the gaps and inform creators on how to make their inventions more accessible. Conversational AI is in this phase and swiftly advancing because of rapid user preference for this kind of user interface.

Users, talking to their mobile phones and marginal conversational agents, are crowdsourcing a massive repository of incredibly diverse training data.

Meanwhile, users and enterprises, over the last couple of decades have sourced and published nearly the total of all human knowledge on billions of websites. The vision of the Library of Alexandria has been realized in the Googleplex. Google (mainly) is indexing, organizing, optimizing, and learning without pause from publishers and searchers. This index has created another massive set of training data.

But, these two activities are not yet ideally interfaced. Web users are still using old interfaces that force them to make compromises to the less conversational computers.

For example, if you go to find something on the Internet today you probably still break your search down into keyword fragments. You’re subconsciously trying to think like the computer to get the most relevant response. Then, you go to a variety of websites and hunt and peck around using a combination of navigation cues and guesses to try and find what you need.

The first process, search via Google, is getting much closer to conversational AI. I can type in a question and typically get a pretty reasonable list of potential answers and often time a direct summarized response. However, the second process, giving users answers to their questions on a website is a marginal experience even on the best-designed sites.

The problem is understandable, a personal blog, small business website, or even an enterprise website doesn’t have enough conversations (traffic and interaction data) to sufficiently train its human designers to understand, process, and generate natural responses to their customers’ visits.

So, the question becomes: Is it necessary to put this burden on websites and their owners? After all, that responsibility exists only because the web browser is the default user interface.

Removing the limitations of current web user interfaces is where Conversational AI starts to get interesting as a superior web UI/UX.

Conversational AI platforms and tools are giving enterprises, and even small businesses access to powerful NLP and machine learning at a technically accessible level for the average programmer or even business analyst. As this challenge to deploy bar increasingly lowers, adding this conversational layer to your website or application is going to become expected.

The added benefit to the business is also compelling. If I can converse with your website — and it works — then you can focus more resources on producing valuable content and less on organizing it for customer discovery.

How Conversational AI will make the web disappear

It’s already happening.

Navigation is disappearing from websites, reduced in many cases to a hamburger menu with a few obligatory links (i.e., contact and about pages). Websites, and the people that design them seem to be assuming that Google search is the user’s preferred navigation, especially for deep pages on our site.

Think about your behavior.

Here are a couple of the typical behaviors I witness in analyzing people asked to navigate the web.

First, and probably the most common is a user asks Siri or Google Now a question in hopes of getting a direct answer or at least a relevant list of web pages that might help.

Second, if I happen to be sitting in front of a desktop or laptop, I’ll type that same question into the search bar (formally known as the URL bar — how fast we evolve) in my favorite browser, which, interesting enough, will produce the same result. Google or Bing attempts to give me a direct answer along with a list of other possible direct answers, along with a relevant list of web pages.

Behind each of these scenarios is Conversational AI.

The approaching reality is that web users don’t need our weak attempts to anticipate their needs with human-crafted user interfaces and experiences. Very soon, it will be less critical to design navigation and intuitively organize our websites. Instead, we will just load up great content and let AI agents arrange, sort, interpret and talk to the humans.

Before I leave you and to tease my next article, could web designers evolve into the intelligent designers of Conversational AI experiences? Think the eerie hidden corridors of Westworld.

Further Reading


If you enjoyed this article follow me on Twitter @billrice and let me know your thoughts?

Also, if you’re looking for an agency that is willing to dig in and understand bleeding-edge technology and get your startup’s message out in the market — Hit me up at bill.rice@kaleidico.com or visit Kaleidico.com.

Photo by Kevin Ku on Unsplash