AI chatbots have come a long way, but are they ready to take over to fill a role in tech support? Jack Wallen addresses this looming question.
Way back in 2016 (because time moves so much faster in tech than the real world), Google's CEO, Sundar Pichai, was quoted saying:
"Looking to the future, the next big step will be for the very concept of the 'device' to fade away...Over time, the computer itself — whatever its form factor — will be an intelligent assistant helping you through your day. We will move from mobile first to an AI first world."
Between then and now, Google first released their Assistant platform through Allo and then used it to re-envision Google Now, such that the platform's digital assistant was more of an AI ecosystem. And it works quite well.
But then, while Google was introducing AI as a mobile assistant (and Siri was continuing to grow more and more intelligent) something interesting began to arise: AI bots as tech support. This first came to my attention in the form of Ask Wiz. Although not actually driven by AI (but by a network of humans), I believe the goal with that app is to eventually have enough Q&A in the database to migrate it to an AI-driven platform.
Ask Wiz isn't the only game in town of course. Do a quick search and you'll find companies, such as Neva, who have developed AI-driven tech support platforms. And then there's Flow.ai, or any given chatbot technology that would empower a company to create AI bots to assist clients and customers troubleshoot their software/hardware/systems.
That is the big, important question, correct? Why would you want to employ AI (or a botnet) for tech support, especially when the standard model works so well? Let me answer the question from two points of view:
- Tech support specialist
- Company owner
I spent a number of years as a remote tech support specialist. I worked the phone all day, answering questions and solving problems. Of the thousands of desktops that were supported, you may (or may not) be surprised to find out the nature of support calls most often consisted of:
- Outlook issues
- Printer problems
- Wireless connectivity
With the vast amount of clients/customers, over 75% of my work involved the above three topics. There'd be days where I would deal with the same issue, call after call after call. That, my friends, is a perfect condition for the AI chatbot tech support solution. A company could generate AI chatbots for printer issues, for network connectivity, for Microsoft products, etc. A client would only have to point their browser to the tech support site, and start chatting with the bot. Of course, without network connectivity, that solution becomes problematic (but there's always a mobile device within reach).
Next we have to take into consideration the point of view from the bottom line—I mean, the owner. Consider the above scenario; the company is paying a support tech to solve the same problems over and over and over and over. The idea of replacing that support tech with a chatbot holds a fairly significant appeal (especially when a company can make use of open source tools, such as ChatScript or the cost-effective, proprietary solution, Dexter to create their AI chatbots). Although building such a system could take significant time and resources, the end result could bring about a considerable cost savings for the company.
Sure, 75% of my time as a support tech was spent on three issues; but when that other 25% rolls around, I had to have the knowledge and skill to take care of those questions and lead those customers back to productivity. Could a chatbot AI, programmed for a limited scope, handle this? No. Neither could a chatbot develop a rapport with clients. By creating a friendly and familiar atmosphere, you make it such that clients are less likely to call in with a sense of pre-established frustration. Yes, they are facing yet another technical issue, but they have developed a working relationship with the tech support specialist and can place the call with next to no trepidation. With an AI chatbot, that is not the case.
Another ding against the chatbot solution is as simple as it is complex. Creativity. There were many, many situations (as a support specialist) where I would have to employ a modicum of creativity to get an issue resolved. Would an AI chatbot be capable of such tricks? No. Chatbots have their scripts and they cannot extend beyond them (at least not in their current iterations). If those chatbots were true AI, all bets are off.
A long way to go
Back in 2014, a Chinese girl named Xiaoice was interviewed (via the social networking platform, Weibo) by journalist Liu Jun for the Chinese newspaper Southern Weekly. The interview was conducted online and Liu Jun had no idea Xiaoice was not human. Xiaoice was a chatbot, developed by Microsoft and eventually was rolled out to lead to one of the largest Turing tests in history.
However, not one chatbot (including Eugene Goostman) has ever passed the Turing test. To that end, we have a long way to go before AI chatbots can truly take the place of a living human for the role of support technician. Even if you only consider the need for creative approaches to solve technical problems and the necessary rapport between technician and client, it becomes instantly clear that AI chatbots are not quite ready for prime time. Will they ever be? If I had to venture a guess, I'd say this is really not a question of if, but when. The AI chatbot will eventually take over as tech support (when and where applicable). As to the when? That's a question for the likes of Chomsky.
- Your advertising stinks. Chatbots can help (TechRepublic)
- Microsoft to devs: Here's a bigger toolkit for adding AI to your apps (TechRepublic)
- How AI will dominate the future of banking, and what businesses can do to catch up (TechRepublic)
- How to build your own bot with Dexter (TechRepublic)
- Next Insurance launches Facebook Messenger chatbot to replace the insurance agent (ZDNet)