Making Conversational Interfaces More Human
In our previous blog post we had discussed about the evolution of contextual conversational interfaces and why they would be something that companies would need to consider for their operations. But after reading through all of that, you would think that it would be easy for you to go out and find a solution to help your organization. This was exactly the fallacy that led to everyone hoping for chatbots would be their savior in 2016, with them being implemented in every possible way. Let us dig into why this happened and what to keep an eye out for.
In 2016 Microsoft CEO Satya Nadella proclaimed that “bots are the new apps”. This followed Facebook opening up its messenger interface for developers to use. This essentially led to a gold rush with every company trying its hand at a chatbot technology. People assumed that they could depend on them to answer just about every question their customers would pose to the company thus, saving them time and effort.
Most of the conversational interfaces available currently tend to be very specific for a particular usage. They perform well as long as users stay within the boundaries drawn by their developers. Though there were some success stories when it comes to chatbot implementations in some sectors at that time such as in Banking and Law practice, they were few and far between. What these companies were able to leverage was the fact that if you created the chatbots to work on repetitive tasks, they excelled. The expectations quickly came down crashing when companies realized that the benefits were overstated if the solution was not developed in a focused manner. The bots could not handle all the complex questions and queries that were posed to them. Thus, in essence, chatbots were good at dealing with limited quantities of information.
But do not be disillusioned! The technology and the thought process behind development has advanced significantly since then.
This shock of a failure helped shine spotlight on different aspects of bots that was never considered when the tech companies pushed forward with their foray into this endeavor. Renowned MIT professor and serial inventor Alan Kay has said “Context is worth 80 IQ points”. Creating a chatbot capable of understanding context and responding to unclear queries has proved to be much more challenging. There are three key aspects that need to be considered to develop such contextual bots.
Natural Language Processing (NLP)
NLP is in essence how the computer understands our language. You would think that computers have been able to process natural language from the beginning, but you would be mistaken. Computer code will only compile and run if it is hundred percent free of spelling and syntax errors. Sounds simple but think about how much time you had spent as a child learning a language. Does not sound as simple anymore, does it? Understanding natural language is difficult because it is messy. Natural languages are composed of large diverse vocabularies, words that have several different meanings and this is not taking into consideration the speech that has different accents and people making ambiguous sentences. We’re not even talking about grammatically formed statements yet. We as humans are able to understand and comprehend a sentence with significant ease even if the sentence has bad grammar or in some cases even wrong spelling. This example makes two things abundantly clear: first, how amazing a human mind is, that it’s able to comprehend such a complex task and second, how complex a process it is to train a computer to understand human language. NLP has seen significant advances over the years. Take for example the word “foot”, it can be both a unit of measure and a human body part. The training of the computer to be able to differentiate between the two is where NLP comes in.
NLP does not help computers read text the same way as humans do. What it does is to help the computers find patters and these patterns coupled with their repetition help the computers identify that something is going on in that sentence. Recognizing the user’s intent using NLP techniques, personalizing the conversation based on the user’s profile and behaviors and using advanced machine learning techniques to continuously improve the process are how bots are becoming intelligent in this space.
Data Processing and Organization
Once the data has been reorganized by the program using NLP, the next important step is to find the relevant information from the enterprise information and data. Data/Information integration and organization is not a trivial task and needs to be carefully curated and deployed. The data that is processed and stored can be further leveraged by applying the different machine learning techniques to further refine the results. The same information can also be used to optimize the conversational flow on the user’s front end. Take an example of the voice assistants on your phone by Apple in the form of Siri. Based on the questions posed to it, the product has continued to develop and to better answer customer needs. If proper processes are employed by the enterprise offering the solution, it should make you more confident that the investment that you are making is not a short term one but a long term one, which will provide significant benefits.
All this needs to be taken into account along with the fact that a customer will communicate with you via multiple channels. In a world where omnichannel is becoming ever popular, customers can use multiple channels — emails, chatbots, messengers and customer service agents — to interact with organizations, and they expect a seamless experience as they switch from one channel to another. A conversational interface that understands how customers use and experience the different channels is critical to deliver a personalized customer experience. This is why one needs to be on the lookout for a product that has effective data processing and organization capabilities that will help the conversational interface to maintain context across conversations with the same customer.
Natural Language Generation
We had explained earlier how difficult a task it is to train a computer on understanding natural language, now imagine how much more difficult it will be to get it to respond in natural language i.e. the reverse. This is possibly the most important aspect as this involves dialogue management in terms of either asking further questions to get the context clearly or providing the answer in a proper sentence. Now if you are considering investing in chatbots you are trying to leverage this technology to help your organization. What you cannot forget is that your organization is dependent on the customers. No customer wants to feel like they are being assisted by the Terminator T-1000 robot. As the number of computers that are surrounding us increase, customers are putting a premium on good service. You want to invest in a service that can make your customer feel at ease when they engage with them.
So, does all of this make you wonder if it’s all worth it? Why go through the trouble when it’s easier to just maintain the status quo? In the post-pandemic world, customers can be expected to engage with organizations at any time, through any channel, and on a self-service mode. Anything less will frustrate your customer. To fulfil these needs while keeping your costs down, it is essential for you to have a formal organization-wide conversational interface strategy. A chatbot is not something that needs to be considered as a standalone but one that will draw on the entire organization knowledge to offer a seamless experience to your customer so you can reap the benefits.
We, at Prescience, are thinking ahead and creating a conversational interface product that takes all of this into consideration. We, as a company, believe that humans deserve better. Keeping ourselves abreast of latest innovations and incorporating them into our products is what we excel at. This translates to our products, with us crafting products that help organizations.
Stay tuned to know more about our latest innovation and how we can help your organization grow and be more productive
Shiva is a keen follower of scientific trends and is an Asimov fan. Believes solid execution is key to the success of any strategy and is focused on building a world class data science team at Prescience. Has a B.Tech from IIT Delhi and MBA from IIM Lucknow with 20+ years of experience in the technology space.