Does a chatbot need to impersonate humans? Is it not enough for it to quickly deliver responses that speak directly to human needs? To boost the business outcomes and deliver superior experience they must be able to deliver quick responses and continuously learn to apply meaningful responses for their unique requirements over time.
Chatbots have started taking over the world, technologically – helping the humans to interact with the machines in their native language. The chatbots have started gaining popularity in many industrial scenarios – particularly the customer service. The tremendous growth of messaging apps, the advancements in artificial intelligence (AI) and cognitive technologies and a wider reach of automation in various processes and business operations are driving force behind the chatbot trend.
Is this all sustainable over time? Well, the factors are making the interest in chatbots growing over time but for it to be sustainable for long time, a stronger business rationale and better near time results will play an important role.
The main challenge faced by developers is to imbibe natural, human like voice in the chatbots so they can converse in the users’ language. Although there is advancement in AI but still it’s not able to fulfill the promise made by the natural language processing (NLP). Most of the chatbots work on deep learning models which are more retrieval based instead of generative ones; making it easy for the users to know that they are talking to a machine, not a human.
Also, in some time, chatbot novelty will not exist anymore. This will leave the users with one concern, how well the bot can get the things done. For keeping the chatbots effective, the best way is by keeping the simple design and minimum conversation. This can be achieved very well by developing utility bots that specialize in specific tasks, provide specific recommendations to the users and help users to complete those tasks.
Bots are being used to perform rote and repetitive tasks but with over time they will evolve to learn and offer personalized interactions with relevant recommendations. This would be achieved by their ability to access and process the data quickly using technologies like neural networks and machine learning.
Some of the best things which can help these machines o comprehend large chunks of data are the self-learning algorithms and graph databases. They make it easier for the machines to understand what the users are referring to and that too without the ambiguity. However it’s very important to dynamically update these semantic graphs which store the data types and their relationships.
The chatbots which are effective will demonstrate and understanding of user needs. It will then complement these needs with visual aids like quick access buttons and images that depict the options available. This way, the chatbots can reduce the user’s time and effort spent on interacting with the chatbot leading to a quality user experience.
The use of chatbots:
The chatbot ecosystem is shaping up; we can see specialized chatbots that are emerging to address various user needs. On the launch of Messenger, Facebook has reported development of over 33,000 bots that too in just few months of the launch. Some of the ways of using the chatbots:
- Customer service – Most of the bots are providing shopping assistance and product recommendations based on user’s preferences.
- Scheduling – The chatbots can be used to buy plane tickets, booking movie tickets and setting up meetings
- Status Checks – These include the news, weather forecasts, local events and everything you need from the internet. The apps/chatbots are there to help you out with these.
- Entertainment purposes – Need random quotes or jokes or funny videos? Chatbots can help the users with these as well.
Whichever way they are being used; it’s important to examine the growth of this phenomenon. It will help in knowing that whether the chatbot will continue the same way or evolve into something entirely different.
The chatbot growth:
Since the evolution with the chatbots and advent of digital revolution, the chatbots have moved mainstream. The movement started in 1950 with Alan Turing’s intelligent machine, later on followed by ELIZA which was intelligent machine to perform near human conversations.
The chatbots have been growing and this can be attributed to number of factors:
- Rising use of Messaging applications among younger generations – One by one, messaging apps have been surpassing the usage of social networks. WhatsApp hit the one billion user mark in 2015. Messaging has been seen as the single biggest contributor to the chatbot growth over the years. Messaging apps are now the platforms and a phone is a messaging device.
- High app volumes – Users are no doubt overwhelmed by app volumes. This doesn’t mea that the apps will give way to chatbots; however there are few apps which users use on daily basis.
- The advancement in AI and related technologies – Artificial Intelligence (AI) industry including machine learning, natural language processing, image recognition and speech recognition is expected to reach a whopping $5.05 billion by the year 2020 growing with a CAGR of 53.65% from year 2015 to 2010.
- Focusing on Overall Customer Experience – Businesses are trying new ways to satisfy the customers on the phone and lower the costs involved. Some of the simple tasks can be done with minimal effort and customers find it annoying to call the contact centre for the same.
- Wider reach of mobile and messenger apps in developing countries – Developing nations like India, the mobile apps and commerce have gone ahead of conventional e-commerce. In many villages and small towns with low bandwidth availability, messenger apps are used for communication, collaboration and commerce. In this scenario, chatbots designed specifically to address these conditions will push their growth in these emerging economies.
All this has been made possible by Digital World. Digital makes 24X7 availability and engagement possible. Digital technologies provide the required infrastructure for these chatbots to live and prosper.
All the above discussion leads to one question, how the chatbot needs to be designed?
The Concern over Sustainability:
The technology today focuses primarily on enhancing features and improving the functionality to make the chatbots engage in a more natural and intelligent conversations like a human responds. All the efforts being put into chatbot to make it behave like a human, it’s success does not wholly depend on its ability to behave in a natural, human like way.
It’s a fact that text messaging is around since last two decades and commercial SMS were introduced in 1990’s. Social boom hit a few years later into the technology world. This led to the inception of the buzzwords – social sharing and enterprise social. It was the time since when the virtual chatting has emerged as a way to build relationships. This is however; set to wear off with the chatbot novelty after which such interactions will not be about building the relationships.
Several attempts have been made in the case of chatbot. The attempt to imbibe emotions such as remorse, happiness, pride or frustration can lead to an experience which is negative for the users who attempt to work with such chatbots and/or machines that try to mimic humans and their behaviors. Also, the attempts that are made to humanize the chatbots can give rise to a term ‘the uncanny valley’, the theory that humans’ emotional response to a robot quickly turns negative when the robot appears ‘almost’ human. The same is applicable with the chatbots as they swing between being intelligent and acting as a human being.
An example would be the Microsoft’s intelligent bot Tay who was criticized for posting racist tweets on the popular social network just a day after its launch post which the company suspended the bot to make the needed refinements. Humans are very much capable of sensing when they are interacting with a machine or chatbot and any of the attempts to make it sound more human like rather than intelligent will not lead the expected outcomes.
One more consideration which is needed to be addressed is the ability of the chatbot to handle complex problems when it comes to customer service environment. The chatbot is very useful when the problems are well defined, clear, easy to articulate and straightforward to solve. Problems which are difficult to explain a human over the other end of the phone/computer are even more difficult to explain to a machine/chatbot on the other end of the phone/computer. Chatbots, in most cases are trained to ask the users to repeat the questions, rephrase it or simply to try again. If it asks to repeat again and again, it can be a frustrating experience for the user.
Utility and Personalization – Two Values of the Chatbots
There are many gadgets which don’t pretend to be intelligent. They don’t engage users. They simply do what they are designed to do, for example coffee vending machines in the offices, ATMs in their initial days at al.
Chatbots are expected to mature along the following lines over time:
- They will be more ‘utilitarian’: They will be very useful for the day to day activities. Users will not hesitate to interact with the chatbots on daily basis.
- Key Performance Metric would be first-contact resolution: The chatbot which would be able to resolve the problems, even the complex ones, without rephrasing or asking the user to repeat the question will lead the chatbot world. The chatbot’s ability to respond to queries in the shortest amount of time and the best way possible will further enhance its utility.
- Understanding the individual user over a period of time, learning the nuances of the user’s request and improving the first-contact resolution: Chatbots which would be able to personalize responses based on user’s query will determine the quality of larger customer experience. A relationship building ability of the chatbot with the user based on their individual needs rather than providing the generalized set of responses will be the trait of a successful chatbot.
- Ability of the chatbot to leverage the data ecosystem – to pull out the data, both from within and outside the enterprise: The faster the chatbot accesses and processes the data, way better will be the user experience be.
- Chatbots will be more specialized in the coming times: Chatbots would be developed for specified tasks rather than being jack of all trades. However, it will take trial and error to know what task it can do perfectly.
The best chatbot experience is when the human get what he or she has requested.
Designing the chatbot – Key Elements
It’s essential to design the bots or the coming times. This means designing the chatbot to work in today’s context. This involves several factors such as bot/human interaction and interface design.
With this, following are the key design elements when it comes to designing the chatbots:
- Reduced manual effort: The best experience which would be key design element would be reduction in manual effort required for human interaction. This will essentially mean, to reduce the number of mouse clicks, keystrokes or touches required the chatbot to determine the best solution to the problem.
- Predicting the option which is right: The system should be able to display the right clickable options for users to choose from, from the available set of choices. This means that the system should be able to understand the problem of the user well enough.
- Personalization: How much personalization can be brought into the bot/human interaction itself? The system must be capable of remembering user profiles, previous conversations, profiles and interactions of the other users in the system, the current context of the problem and the environmental know how.
- Best approach for the unresolved queries: When the chatbot is unable to understand or resolve a particular query or user is not able to phrase it as expected, the level of frustration would be higher. In this case, best design strategy would be that the chatbot should demonstrate a partial understanding to reveal which part of the question the chatbot failed to understand.
- User Recommendations: In the case when the conversation is stuck in a loop with no resolution in sight, one resolution is to provide recommendations for the users to bring the chatbot into more useful state.
Future of the chatbot, in a nutshell
Most of the bots today have failed to meet the expectations around them. Blame it to the slow advancements in the technology. It’s fairly easy to design and deploy the chatbots today. Businesses are developing the chatbots to reach people where they are spending their most of the time, like messaging apps such as WeChat, Facebook Messenger at al.
However, there are other ways of reaching users on messaging apps and WeChat is just one primary example. Developers must be motivated to assemble experiences with bots at the center of the action rather than just a technological adventure.
We are building a platform where you can create a custom chatbot and publish it on Facebook, Telegram, Andriod and iOS. Kindly share this article and follow me on Twitter.