Chatbots: whole lotta hype - TechCentral

Chatbots: whole lotta hype

I have a soft spot for helpdesk chat services. I always try my Internet service provider’s chat client before anything else, for three reasons: it is far cheaper than a phone call, it’s much faster than an e-mail response and it allows me to juggle other tasks while finding out why my Internet isn’t working.

Most recently I used the chat client on my mobile provider’s website for the specifics of a deal I was interested in. Overall, chat clients are very useful as far as customer interactions go and they are a very welcome part of the experience.

But we are a long way away from these channels becoming automated, at least judging by the questions I ask. For example, I’d never bother using a chat client to find out a company’s address or where I might find a driver download. A Web search is much better at that. In fact, I never use the client unless I have a topic that requires proper answers from someone who grasps the nuances of my concerns.

Bots don’t ‘get’ people

A popular podcast recently looked at the rise of machines and whether they pose a threat to workers. This gaze of automated unemployment often falls on the service industry, yet that group is possibly the least concerned about being out of a job. Repeatedly, every interviewee on the podcast said that computers have no chance dealing with the irrational nature of customers.

I’m inclined to agree. When I do want to speak to a human, I usually have a situation that no automated information system (such as a website) is able to resolve. In fact, when I visited the mobile store for that deal I mentioned, right next to me sat a woman deeply unhappy about something. She was trying to cancel a contract but I think she was in arrears. Regardless, no matter how the representative tried to explain the situation to her, she was stuck on her own narrative.

Sadly, her problem was not resolved and she left in a huff. Yet if she had to dance that dance with a machine, the outcome would have been far worse. There is no way an algorithm would have been able to grasp the nuances of her anger. Even a tired rep was at least intuitive enough not to piss her off completely.

These thoughts mingle in my head as I try to understand the movement of chatbots. Have no doubt: they are a growing force in our society. The recent Gartner Symposium held in Cape Town yet again raised chatbots (now split into virtual personal assistants, consumer personal assistants and employee personal assistants).

Chatbots have become aligned with the chatter about artificial intelligence, a topic that is about as well understood in general terms as quantum mechanics. As such, chatbots are being treated with a certain mysticism. The Gartner presentations didn’t say this, but the analyst firm did note there is a significant and misunderstood difference between the proxy that lives on your phone or on a website, and the cloud-powered brains that makes the magic happen. This may change drastically with Intel’s newly announced AI chip, but that is a topic for another day.

The hype around chatbots is not about their intelligence or that they may one day be as good as a bored sales assistant at handling irate customers with strange requests (or who often only want to be heard out.) The hype is due to an industry hell-bent on making chatbots the next big thing. Here is how they do it.

First, they draw comparisons to the booming chat app market, because this is the place chatbots will most likely thrive. They then point to slowing app numbers, particularly social apps, and draw a disingenuous causation link between the two: talky apps are up, other apps are down, so surely the market is going towards the talky stuff.

App decline

The chatbot hype wagon is securely fixed to the perception of app decline, in that it will fill this alleged vacuum. Then they finally make a big leap: Company X has a messenger app and it has bots, so all it must do is monetise the platform and make billions! Yes, and all I need to reach the moon is to don a pressurised suit and get into a rocket ship.

Look closely and you find these reports are full of words such as “should” and “could”, neatly wrapped in the ambiguous and paranoia-inducing term “customer experience” (and other vague buzzwords such as “less friction” and “streamlining”). They offer hardly any hard data that chatbots are actually making huge gains for whatever business case they cover.

I’m sure chatbots have been useful to move a few more movie tickets, catch a handful of pizza deliveries or act as a glorified RSS feed. But these are hardly blips and not worthy of the grand aspirations being cast towards bots. If they were, we’d see the numbers. Both Pizza Hut and Domino’s Pizza have had order bots for over a year, yet you won’t yet find mention of their impact.

“If all these bots can do is order pizza and give me search results, they are not going to be of much use,” says the author, James Francis

Indeed, the Gartner presentation pointed out a key fact: the chatbot market is so Wild West that there is no best practice. A further observation from me: almost all chatbots are inside walled gardens. Facebook’s chatbots only work on Facebook’s platforms. If my ISP decided to use a chatbot, it would likely live on its site. I can’t send a WeChat message to the Facebook bot and I can’t interact with my ISP’s bot through WhatsApp. This is rather limiting, like a browser that only visits one company’s sites. Though advances are happening here, chatbots are still more digital fiefdoms than communication liberators.

Gartner noted that bots that chat to bots may be a solution to this. So I might have a personal chatbot that talks to other bots. Of course, we still live in a world where I can’t even get Siri to interact with most of my phone’s apps, let alone handle relatively complex chains of commands.

Chatbots are more hype than anything else. It is smart to experiment with them. But we’ve yet to see the market-defining breakthrough that will make these systems appeal universally. Right now, they are a victim of innovator’s determination and speculative market analysis.

For example, the CNN Facebook bot is just a dumb aggregator that doesn’t seem to do much than spam headlines at me. And the Alex WikiMessenger (also on Facebook) appears to just be a glorified way to search Wikipedia.

If all these bots can do is order pizza and give me search results, they are not going to be of much use. Chatbots are certainly part of our future. But they need to do a lot more than they do now – like telling a bot to “call TechCentral, tell them the column will be late, and then send then a gift basket to make up for that”.

Now that will be awesome (and x.ai’s Amy is a step towards that future.) But current chatbots appear about as close as the Apple II was to the iPhone.

  • James Francis is a freelance writer whose work has appeared in several local and international publications

Comments are closed.