Artificial Intelligence

Chatbots are dead. A lack of AI killed them.

Chatbots were all the rage back in 2017, but in 2018 we're still waiting for them to live up to their hype.

If a chatbot falls alone in a forest, does it make a sound? Given that basically no one uses chatbots, does it even matter?

Chatbots were all the rage with the media but never caught on with developers or the consumers they serve. As Digit's Ethan Bloch has eulogized, "I'm not even sure if we can say 'chatbots are dead,' because I don't even know if they were ever alive."

No, really, they're going to be huge!

While Bloch is right to say that "No one can point to a chatbot that 'all your friends were using'" as "Such a thing simply never existed," once upon a time many pundits pointed to chatbots as the future of commerce, social, and just about everything else. Chatbots were one of the big themes of Mobile World Congress 2017, with the conference organizers summarizing the buzz from main stage and hallway conversations thus: "There was overwhelming acceptance at the event of the inevitable shift of focus for brands and corporates to chatbots (often referred to as 'conversational commerce'), reflecting the need for brands to go where consumers are, even if many companies remain uncertain at this stage of the eventual outcome."

While they went on to acknowledge that "the true potential of chatbots will require further advances in AI and machine learning," the people behind the industry's biggest mobile event felt the only significant question around chatbots had to do with who would dominate, and not whether chatbots would take off: "Will a single platform emerge to dominate the chatbot and personal assistant ecosystem?"

SEE: IT leader's guide to the future of artificial intelligence (Tech Pro Research)

A year later, we know the answer to that question: No. Not only will a single platform not emerge to dominate the chatbot ecosystem, no chatbot ecosystem has emerged at all for someone to dominate.

Missing the point

So what happened? According to Dave Feldman, Vice President of Product Design at Heap, chatbots didn't just take on one difficult problem and fail: They took on several and failed at each one. As he detailed in a post mortem on chatbots, several key factors influenced the non-rise (and hence, non-fall) of chatbots:

  1. Platforms are hard (Standards were non-existent and tooling was a chaotic stew without clear guidance on what or how developers were supposed to build useful services);
  2. Replacing apps is hard ("Bots weren't really going to replace apps, any more than apps replaced the web," he writes, and pretending otherwise simply dissuaded developers from bothering with overcooked promises of chatbots' potential); and
  3. Text is hard ("There are technical and UX problems that limit the efficacy of a text-based, conversational UI," he said, and not nearly enough horsepower in the AI/ML powering the understanding of text to usefully "talk" to machines).

SEE: How to implement AI and machine learning (ZDNet special report) | Download the report as a PDF (TechRepublic)

Implicit or explicit in each of these problems is a failure to capture the attention and creativity of developers. For those paying attention, this was an obvious gap, and one that I called out in 2016:

According to the VisionMobile developer survey, a mere 4% of developers are actively building chatbots today. That number jumps to 20% of developers indicating they plan to develop chatbots within the next 12 months. That's a significant number, to be sure, but becomes less so when we realize that fewer than 25% of developers that are aware of chatbots (86%) are convinced that they're worth their time.

As I argued then, "To get developers interested, the early chatbot developers, and particularly the biggest ones—Facebook and Microsoft—must demonstrate considerable interest from users," which sounds simple until you realize that consumer interest was never going to materialize until machine intelligence could get anywhere near human intelligence. User interest, I said, "depends upon AI that makes talking with a bot worthwhile for consumers." For those of us that tinkered with chatbots early on (and even later on), talking with bots yielded precious little value.

Which is where we are today.

Feldman believes there's a future for chatbots as additions to apps, web, and other experiences. This is probably right. But to get there, we not only have to scale down our expectations for chatbots, we basically have to dismantle the very definition of a chatbot. What we used to define as a full-fledged, intelligent agent has become a light chat-style interface bolted onto a webpage (for customer service use cases, for example). It's not sexy, but maybe unsexy is the future when it comes to chatbots.

Also see

chatbot.jpg
Image: wutwhanfoto, Getty Images/iStockphoto

About Matt Asay

Matt Asay is a veteran technology columnist who has written for CNET, ReadWrite, and other tech media. Asay has also held a variety of executive roles with leading mobile and big data software companies.

Editor's Picks

Free Newsletters, In your Inbox