Genie in a Bot: Support over Chat
You could blame this on pop culture, but every time I think about Artificial Intelligence (AI), my head fills with images of a killer cyborg army. However, recent advances in the field of software intelligence paint a completely different picture. Contrary to apocalyptic visions of artificial intelligence, IBM Watson actually helped save a Japanese woman’s life earlier this year. When she wasn’t responding to treatment for leukemia, doctors sought the help of Watson. Within 10 minutes, it had scanned over 20 million clinical oncological studies and identified an exceedingly rare form of the disease.
AI-lementary, my dear Watson.
Think about that for a second. 20 million clinical oncological studies! No human can even dream of taking up the task, let alone have a chance of completing it on time. A woman lives because of these technological advances in software, and the prospect of how AI is going to grow from this point is really exciting.
Of course, what can be used to save lives can also be used to improve standard of living, whether it’s self-driving cars or chatbots in customer service. Over $300 million in venture capital was invested in Artificial Intelligence based startups in 2014 to help make customer service a more efficient and painless process. The rise of chatbots and smart replies are the clearest indication of AI improving the standards of customer service. Confused? Let me explain.
A Tale of Two Bots
We grade the intelligence of a software based on how well it can mimic a human being. If you can’t distinguish between a human being’s reply and a computer’s, the computer is considered intelligent.
The sign of a truly intelligent system is when, a bot is able to indistinguishably mimic the conversational pattern of a human.
In today’s world, a chatbot that sounds like a human being belongs to one of two categories:
Remember the last time you had to sit through that long, long automated call punching 2 for English, # for the main menu and so on? Well, a rule-based chatbot operates on a mechanism similar to these Interactive Voice Response calls. When you begin a chat, the chatbot presents you with a list of options, you pick one, the chatbot responds with a pre-programmed answer and the conversation flows in a structured manner.
Although rule-based chatbots are really efficient for collecting basic information or routing questions to specific departments, these interactions mostly end with the person speaking to another human anyway. The chatbot cannot actually figure out what your problem is and find you a solution for it. Issues that would require some back-and-forth cannot be resolved.
Machine Learning Based Chatbots
ML-based chatbots fill in the gaps left by traditional rule-based chatbots. ML-enabled bots can identify the intent of the customer from conversation, and then take the user through a decision tree, a flowchart or, even provide options/solutions to users.
Based on its understanding of the scenario, it will infer context, refer to past events and offer resources to help with queries. The best part? It’s always learning. This means that an ML-enabled chatbot will gradually make better decisions and improve itself, with every interaction.
All this may sound good on paper but what makes them more than just a proof of concept? What application do they have in real-life scenarios? How exactly are they improving the standards of customer support with chatbots?
Chatbots in support
There was an interesting trend towards the end of 2015. The number of daily active users on messaging apps leapfrogged that of social apps. This points to the preference for real-time conversations and its potential in the business world. Well, messaging does reduce friction. Think about it this way: customers just have to open up an app or just click on that handy help button in your app. Where once a customer might have had to write a detailed email, or frame a detailed rage tweet, they are now more likely to send you a message.
Chatbots can be the first touchpoint in a customer interaction and automate things like the collection of basic information, and routing customers to departments. This way, chatbots can conveniently be used to reduce the physical effort taken to reach from point A to point B in the complex customer support circuit.
When the concept of IVRs was introduced, stats saw up to 75% of callers’ issues being resolved even before they reached a live agent. If such miracles can occur with a process as cumbersome as an IVR, imagine the potential of incorporating automation with chat!
In addition to the cases above, chatbots are also capable of a wide array of automations across any industry. For example, if a customer of an e-commerce business is complaining about a damaged delivery, a bot can assess the situation and ask them if they’d prefer a refund or a replacement. Based on their answer, it can initiate the required process on its own. A frequently encountered problem solved, without ever requiring that an agent step in! It’s self-service, bot style.
The chemistry between chat and bots is arguably better than that of the phone channel and IVRs. A large chunk of time spent listening to all the options in an IVR menu could actually be spent looking for answers. Over chat, you can quickly skim through all the options and actually end up knowing more than when you started. Listening to long, complex menus can be pretty tiring.
I don’t know about you but, I’d rather just press 9 and talk to an agent directly, than suffer through a cumbersome and confusing list. And when I find out that 9 doesn’t connect me to an agent, let’s just say I’m not the most pleasant person in the room.
When you’re going to eventually enter your choice on the screen anyway, wouldn’t it be much simpler to just read the options off it instead?
If you want to constantly delight customers, the first step would be to remember they exist. If every time they contact you, your first question is “Tell me your name and email address”, you’ve already lost. You need enough context around them so you can help them with what they need.
So imagine this: a customer has called you 7 times in the past and each call averaged at around 13 minutes. When they call the eighth time, can you really listen to 90 minutes of past calls? A chat log would make this much more efficient. With just a glance, you can see records of past interactions with bots or other agents and assess the customer’s relationship with your business.
Quality of interaction
Chat has a number of advantages over the other channels. You can include attachments, screen grabs, pictures and still send and receive instant, personalized responses. This richness of interaction facilitates a lot of other use cases. Something as simple as pictures could enable the automation of resolving issues.
Take for instance, a customer who has purchased a new router. They’re seeking help to set it up. They get connected to the company’s chatbot which helps them with every step of the process, even attaching pictures when necessary. Pretty neat, right?
With that, I think it would be fair to say that artificial intelligence is swiftly transitioning from the tag of an ‘’exciting prospect’’ to a ‘’blatant eventuality’’. As systems get smarter, and algorithms evolve, modern software will challenge the boundaries of conventional customer support processes.
While the wait time for Artificial Super Intelligence is indefinite, the wait for our next post is relatively shorter. Stay tuned for our next post in this series where we’ll talk about the evolution of Machine Learning over the years.
Also, make sure to leave your thoughts on modern chatbots and the future of AI in customer support in the comments below, or drop us a mail at firstname.lastname@example.org. I promise I’ll read them all; of course, until they replace me with a bot.
“Maybe the only significant difference between a really smart simulation and a human being was the noise they made when you punched them.” – Terry Pratchett, The Long Earth