In 2017, even the most stubborn hermit has encountered (knowingly or otherwise) a chatbot. But what are they, what are they good for and how are they being used today? If you’re interested in having some rumours dispelled and get an introduction to the world of chatbots carry on reading!
The concept of robots speaking with humans using human language has been around for decades and they’ve made appearances in various industries in recent times, but it’s safe to say they’ve failed to properly take root. Today however, many tech (and chatbot) evangelists are preaching that 2017 is the year of chatbots. As per usual with technologies claiming to be the new ‘It’ thing there tends to be a lot of promises being made and information being spread, making it difficult to grasp where the chatbots are and can do today without having to spend significant amounts of time researching the technology.
As it happens, in the last few months I’ve been lucky enough to be one of those people who have gotten to spend time building and researching chatbots, so in the spirit of sharing the knowledge I figured I’d put together a quick article outlining what chatbots are and where they came from, what they can do today and where I believe (and hope) chatbots will move towards in the future.
What is a Chatbot and why should I care?
Chatbots are programmed to converse with humans as if they were speaking to another human rather than communicating with a computer using a coding language. The idea is old (bordering on ancient) and developers have known how to achieve basic versions of this using If-then statements since the 1960s, during which Eliza boggled minds by simulating conversations with humans.
For years however, most didn’t feel comfortable giving chatbots serious responsibilities like having them face outwards for companies as they often got lost in translation (and some still do), however they were still being developed behind the scenes. Many chatbots can be found in games (The Colossal Cave Adventure, Event) but the biggest advancement in the field has been happening in academia or by hobbyists who simply enjoyed experimenting. Since 1990 some of the best chatbots have gathered under one roof annually to compete for the Loebner Prize; a competition inspired by Alan Turing’s seminal paper on whether robots could think (Computing Machinery and Intelligence, 1950). In the competition humans chat to other humans as well as chatbots whilst they try to discern whether their counterpart is a human or a chatbot, and the chatbot which impersonates a human best is crowned the winner (yet no chatbot has successfully tricked all judges for all questions into thinking they’re human: Humans 1 — Robots 0).
The appeal of chatbots in academia and business has multiple aspects, but the academic attraction deserves a paper (or 10) on its own thus this post will focus on the advantages of chatbots in real world applications.
People are starting to take chatbots seriously
Having a chatbot representing or functioning within your organisation provides multiple benefits. A chatbot is the co-worker that never sleeps, never falls ill, never loses their temper, always stays online, and doesn’t add a cent to the bar tab at the office party. They continuously evolve and improve, and they never get bored by repetition or routine which allows their human co-workers to focus on handling more important matters. They are consistent, and one chatbot can carry thousands of conversations at once without mixing them up (Humans 1 — Robots 1). According to Juniper research chatbots will be responsible for costs savings valued over $8 billions annually by 2022, largely due to their aforementioned capabilities. The ability of carrying multiple conversations at once has the added benefit of gathering information about the customers based on the conversations being held and centralising the information immediately. What words most often appear in a negative context (what would the customer like to see improved)? What words often appear in a positive context (what do they like about the product)? What is the most commonly received question (what has not been explained well enough)? With a chatbot employed information about what the customer values, dislikes or would like to see improved can be gathered in a neat spreadsheet automatically. By equipping chatbots with AI based technologies or cognitive intelligence, such as emotional or personality analysis tools further valuable information can be gathered about the customers (this will be expanded on later).
If the benefits are so obvious, why aren’t they mainstream already?
If you’ve read about chatbots before, odds are you’ve seen this Business Insider graph already. But this article is meant to get anyone and everyone up to speed so let’s include it as it highlights something important: nowadays people find themselves spending more time on messaging apps than social networking apps which is largely why chatbots are starting to explode. The reason here is that chatbots can reach users where they’re already hanging out, without requiring users to download and open new apps. And considering the trend it doesn’t look like people will abandon messaging apps anytime soon.
But it’s not only user habits driving the increasing popularity of chatbots, but also the technological advancements being made in the field of artificial intelligence and machine learning, or more specifically within natural language understanding. Again, this is an article (or 10) on its own but for now it suffices to state that new techniques have been developed in recent years which allow computers to extract important features of human speech, such as the intent of a sentence with relative ease.
If you’ve gotten this far and you’re thinking “Cool, but this doesn’t apply to me because I’ve never met a chatbot”, I beg to differ. Most of us have now met chatbots like Siri, Cortana or Alexa that act as digital assistants for users; able to search for answers, access apps and provide sassy replies to existential questions. But these whales of the chatbot-ocean are far from alone.
You had my curiosity but now you have my attention. Show me some chatbots!
Chatbots are popping up like mushrooms in a variety of fields with specific tasks and abilities, ranging from disseminating information to guiding users through complex processes.
Today there are some well thought out chatbots accessible on websites and messaging platforms. Some are used to disseminate information like the AccuWeather chatbot which can share weather forecasts, or the CNN chatbot which is easy to engage with and capable of providing summaries of articles as well as full articles or alerts. A ‘cousin’ of these are the marketing chatbots with some being employed today to market organisation as well as individuals (shout out to Esther Crawford for her clever chatbot trained to answer questions about her resume and experience). Chatbots in marketing are great as they allow organisations to interact with customers with a personal touch as they can learn names, analyse and adapt based on past behaviour, they can send users notifications, all the while the chatbot gathers information about the customers. The benefits of marketing chatbots are also applicable in E-commerce where chatbots can be used for personal engagement, they can be useful for Q&A’s and they can help users find similar products to the ones they’re interested in, like Alexa which is connected to Amazon (a convenience us Australians are yet to enjoy). Alexa, being connected to other apps also shows how useful chatbots can be when it comes to processes previously requiring forms such as ordering food for delivery (dish, delivery address, mode of payment etc). Here lies another strength of chatbots as they can memorise information about the user, so the second time you order pizza from Alexa all you need to do is pick toppings and size whilst Alexa remembers your address and payment information. These advantages carry over into organising travel as well, with SnapTravel being a great example. This chatbot helps users find travels deals and they have a human in the backend who’s capable of taking over if issues arise upon arrival.
SnapTravel chatbot, the 24/7 travel agent
Many (if not most) chatbots are customer facing but chatbots can also serve a purpose within organisations. The slackbot (Slack) is a great example as it’s capable of helping users prioritise conversations whilst also being able to aid organisations with stand-up meetings. Tangowork too is designed for internal use capable of disseminating information, answering queries or act as an event guide. Speaking of events, chatbots can nowadays be put together rather quickly when designed as event guides for conferences or festivals. This is an ideal solution as many events often require the download of a mediocre app which is only useful for a day, whilst the chatbot can be reached through messaging platforms capable of informing users about times, locations and popularity of particular events or artists.
The aforementioned chatbots are all fun and practical but there are also chatbots out there capable of aiding with more complex processes. Chatbots are being increasingly employed in banking (UBANK, Swedbank) as well as insurance (ICICI Lombard, PNB) where they aid users with commonly asked questions as well as product guidance, where users are asked questions and are recommended products based on their answers. It may come as a surprise considering the complexity of such products but according to Guy Goldstein, CEO and Co-founder of Next Insurance, “70% of our customers are buying insurance on their phones,”. With such metrics in mind the appeal of chatbots should make more sense, thus carefully constructed chatbots have been implemented in these fields and have thus far faced a positive reception.
Chatbots are also helping organisations in recruitment with More2, for example, reporting a better quality of applications after having implemented chatbots for the first engagement with potential recruits. Real estate too are employing chatbots (Automabots, Domain) where they help customers with queries or property information which are later often converted into meetings.
It’s becoming increasingly common to find chat boxes in the corner of websites where visitors can ask questions, placing screens between the actors during pre-sales customer engagement. Since users already have the habit in place of engaging via chat on websites, Round Table Apps implemented a chatbot which acts as a concierge capable of answering questions about the organisation, and it can help visitors set up meetings by leaving ‘notes’, or they can ask to speak with a human in which case the chatbot goes into hibernation as a human takes over.
RTA-D3 learns new things every day, though its civility has probably already peaked
Cool, chatbots are programmable parrots but surely, they can’t ‘understand’ us?
Indeed, chatbots cannot think nor truly understand humans but they are getting better at impersonating our abilities and traits via artificial intelligence technologies such as machine learning and emotional analytics. Combining chatbots with emotional analytics are useful for customer support where humans can take over the conversation if the chatbot recognises that the customer is becoming frustrated with the service. Add on a voice-to-text and text-to-voice program onto the customer service chatbot and it can serve the same purpose over the phone, removing the need for the time consuming auto-attendants list of options read-outs.
As chatbots are improving at imitating human abilities, and since many of our favorite science fiction technologies can only be achieved by using chatbots as a stepping stone; more research and development will inevitably go into the field. But today’s chatbots are far from perfect and they aren’t the silver bullet that will put all humans out of our jobs, allowing us to spend our days happily relaxing at home collecting universal basic incomes. But they can certainly aid us in some circumstances. In customer engagement scenarios where many of the received questions are recurring chatbots are fantastic as they can filter out the most common questions, carry multiple conversations at once, meeting customers with the same collected tone and uplifting mood. In times where chatbots cannot answer a question they can then pass over the conversation to a human who’ll resolve the issue and add the newly received question and answer to the chatbot. And here we find another great strength of the chatbot; it’s always evolving and improving. So although they might not be able to improvise an answer to a new question like us humans can whom are armed with general intelligence (Humans 2 — Robots 1) they’ll never be stumped by a question more than once.
So… What does it all amount to?
Whether the future of chatbots and AI will eventually lead humanity into an automated paradise or our own demise (thanks for the nightmares Hollywood) is yet to be determined.
But what we do know is that the chatbot market is on the upswing and it seems to be for real. Business Insider reported that 67% of consumers worldwide used a chatbot for customer support in the past year, and Gartner predicts that 85% of all customer service interactions will be powered by chatbots by 2020. Impressive numbers considering the industry’s adolescence and that best practices are yet to be established. This lack of best practices may feel daunting to some but is in fact an opportunity: the market is still up for grabs. To stay ahead of the curve people and organisation need to get involved sooner rather than later to experiment and explore the potential of chatbots. The market may be there for the taking today but eventually someone somewhere will lay the path which others shall follow, so why not chuck your hat into the ring and give chatbots a go?
P.S. When I started I thought it would be shorter than it ended up being, so if you made it this far kudos, and thanks for reading!