From Eliza To Alexa, The Journey, The Problems And The Future.
A while back, I wrote a piece about chatbots on my LinkedIn, if you’re interested you can read it here.
I read an article by Venture Beat shortly after which talks about the 4 big problems with chatbots, and I wanted to discuss them, and some more thoughts on the views of AI. Before I go into it, it’s important to note that bots aren’t a new thing, they’ve been around a long time. In fact, the first chatbot came about in the 60’s. Don’t believe me? She’s called Eliza, and you can speak to her here (and even change the code if you want to make enhancements, of which there are plenty). So, what are the problems?
1 & 2. They Don’t Solve A Problem And Overcomplicate A Solution
Chatbots don’t solve all problems, but they do solve some. Consider:
· We now have attention spans shorter than goldfish
· Contact volumes are generally rising across all sectors
· Customers are moving towards a 24/7 responsiveness expectation
· Contact methods are expanding rapidly
From a customer experience perspective, they remove two of the biggest frustrations around speaking to a business, the issues of time and connectedness. A chatbot, even pegged against the fastest and most receptive brands in the world, will answer 1000 times faster than a person, being able to respond faster than a human with more connected information regardless of channel. The projects we’ve done with our clients adopting AI reduced human responses on average by 11% while maintaining or increasing CSAT scores, mostly because they only look to solve the problem at hand for customer service, the four issues listed above.
From a complication perspective, all bots I’ve seen so far reduce overall time to resolution, in fact it’s the only concrete reason for their existence. Unless the hook for the customer is so undoubtedly clear, they’ll choose a human every time. From a business perspective, they answer more quickly and accurately and from a customer perspective they give them what they’re looking for in near real time. But to ensure they don’t overcomplicate a solution, it’s important to remember,
Bots are a mediator, not an end point
Complex queries are the realm of the human, at least for now. The problems for customers around time and connectedness can be massively reduced through AI without having detrimental impact on either side of the fence. In the current climate, chatbots should be human augmented to let real people take on the emotional, important or potentially impacting conversations while bots can gather data and answer routine questions.
3. They Don’t Fully Utilise The Platform
Utilisation of platforms is part of a wider problem in customer experience generally. What bots are missing is what I refer to as a ‘glue layer’ (like Gnatta), they can provide a suggested output based on the NLP engine but can’t do anything with it. Just like there are nuances to each channel a bot sits on, there are nuances of metadata about the customer which are often not taken into account. Consider this scenario: someone contacts you about a stock query through email. Their email tells you nothing about them, but it’s linked to the same account they used to set up their Twitter. Their bio states they’re a journalist for a major newspaper, would you treat them differently knowing that? Most likely. A glue layer does exactly that, sticking together not only different methods of communication, but also linking different systems, be it CRM’s, OMS’, WFM’s and fistfuls of other acronyms to provide a seamless, connected experience. Platform utilisation is one piece of the AI puzzle, being omniscient is the real challenge for bots and customer experience more widely.
4. They skipped user testing
I’ve seen many, many examples of badly executed chatbots; of course chatbots should be tested but I don’t think it’s an overarching problem. Most reading this that have built or been involved with a build will have been in the testing phase for months before going live; it’s important to remember the emerging nature of the market and set expectations against it, but I’d argue it’s not a big problem for bots.
So What’s Changed Since Eliza
It’s simple, really; we’re more connected. Very recently we’ve entered an era of screenless hubs through Alexa, Google Home and more recently HomePod (although Siri and other voice recognition have been around for some time). Through these, we can use bots to fulfil tasks without having to sit down and focus on communicating with them solely.
I’d predict that the boom of chatbots in the last 2 years is the beginning of the trend where we have our own personal chatbot companions as an extension of home hubs.
I touched on this briefly in my previous writing, but unlike the time of Eliza in the 60’s, we now have the ability to connect our bots to other systems to perform executable tasks. The IoT has been at this for some time with things like the connected home; I’d say we’re 2 years away from having our own chatbots which will change the way we speak to each other forever. In this not so distant future, bots fulfil monotonous, boring or predictable tasks such as booking the table for this evening’s meal while we gain back time to develop real connections in a way a bot never could.
With the removal of the graphical interface to communicate with bots and the ability to connect them to any channel or data source outside of home hubs, users can stay native to their channel of choice. It’s likely to be one of the biggest shifts in the way we speak to each other since the telephone. The amount of physical communication we do is likely to drop, while the quality of our conversations, physical and non physical, at least in my estimations will improve, with more time to spend creating bonds with our families, friends and loved ones.