Opinion: Meaningful conversations with chatbots? Nah

Photo credit: Couleur.

Human conversations are spontaneous and interactive. Language is also unique. But digital conversations, which mimic the appeal of a face-to-face conversation through a digital device, are increasingly becoming a norm.

The expanding digital universe has only gotten bigger. It is no wonder why Gartner predicts that “by 2020, customers will manage 85 percent of their relationship with the enterprise without interacting with a human.” But a 15-year-old boy who looks at his text messages could ruefully remark, “One day, I will learn to have more meaningful conversations.”

Striking conversations from history

“Hi, I need some urgent help. My credit card has been stolen!” a human said.

“Everything is urgent for humans. People get on my nerves!” replied Parry, a 1972 bot.

“My account number is 171206 […]”

“I know your account number and your card has already been blocked, silly homo sapien!” sneered Parry.

Eliza, a more sober bot, chimed in,“Ignore Mr. Parry, he’s a paranoid schizophrenic. Ha ha ha!”

“OK,” said the puzzled human, not knowing whether to laugh with the machine or not. “How do I handle this subtle cognitive dissonance?” he silently mused.

When greeted with apparently believable personas, our conversations can strike a meaningful chord. Humans are often fooled into thinking they are communicating with a real person. This is the Eliza Effect from the 1966 chatterbot Eliza. The concept for bots isn’t new and still works on the principle “If this, then that.” Parry came in 1972 and was described as “Eliza with an attitude.”

Jabberwacky, A.L.I.C.E., Smarterchild, Watson, Siri, Google Now, Alexa, Cortana, Tay, and Bots for Messenger have led to an evolution. But have they really managed to strike a meaningful conversation? Or are they programmed to laugh out loud (LOL) at an inappropriate moment and even expect humans to laugh along with them?

Today’s bots: making or breaking conversations?

According to a Forrester report, “Bots Aren’t Ready to be Bankers,” today’s bots fall short of customer expectations by a huge margin. This is because the majority of the bots still use comparatively basic machine learning instead of deep AI-like neural networks. There’s no scope for mediocrity with bots as far as security is concerned. Misunderstanding sensitive customer information or treating it with flippancy can cause catastrophic results to a firm’s brand image.

One needs to set the stage by re-engineering back end systems so that futuristic digital offerings can be integrated easily. Integrating with other existing technologies is one of the top challenges marketers are worried about, according to a report by Demandbase.

As Digiday pointed out, “Facebook said it was ‘refocusing’ its use of AI after its bots hit a 70 percent failure rate, meaning bots could only get to 30 percent of requests without some sort of human intervention.” Fashion retailer Everlane, a pioneering Facebook Messenger partner, has also dropped their bot and will use email instead.

Stitching the conversational thread