Prescience Decision Solutions

Prescience Chatbots NLP Blog

A Rapid Primer for Conversational Interfaces

Since Allen Turing set out with the Turing test to determine if a computer is capable of thinking like a human, mankind has been on a pursuit of what has been called the ‘holy grail’. The holy grail of a human-like conversational interface with computers – a way to automate the responses so as to ease burden on people and businesses. It was sixteen years before the Eliza chatbot was created at MIT in 1966. It used pattern matching and substitution to create an impression of understanding but lacked any ability to contextualize events. This was similar to an “if then” logic to enable a flow for the conversation.

Today, we are naturally accustomed to interacting with our mobile phones via voice interfaces such as Siri, Google Assistant, Alexa and many more. The enterprises that have created such ‘assistants’ are trying to implement Artificial Intelligence (AI) into their interfaces so as to provide more natural and contextual responses.

This premise brings us to the two forms of conversational interfaces with computers:

  • Text
  • Voice

We see the text-based ones in pretty much every website at this point in the form of pop-up chats on websites. In the case of voice-based ones, an example we have all been exposed to is that of voice based interactive call systems used by companies. As these different systems continue to penetrate into our lives but have we ever wondered why we are seeing an increased adoption of such interfaces? Have you wondered if it is a good idea to consider investing in such a solution for your business?

Why this technology is great for businesses?

  • 24/7 support
    In the ever-connected world that we live in, the customers expect enterprises to be available to them at all times. This puts a significant strain on businesses to ensure that every customer query is resolved to a satisfactory conclusion. With this in mind, businesses are increasingly looking for ways to provide such a higher degree of customer experience.
  • Empowering customers
    Customers now are increasingly looking to help themselves via researching and looking up solutions to their problems. A contextual conversation interface helps speed up the process in this endeavor. This not only reduces the strain on the company to provide support to queries but also leads to increased customer engagement.
  • Lower costs
    The one resource that every industry is always looking to conserve, and grow, is its monetary one. More often than not, companies look into implementing such a solution for cost-cutting purposes. The hope is that by empowering customers to help themselves, they can concentrate on continuing to develop products and services.
  • Reputation of the business
    Something that is not taken into consideration often is the importance of a satisfied customer. In our interconnected world, even a single unsatisfied review can cause reputational harm to the company. The internet makes it possible to put a spotlight on a bad experience very easily. This, coupled with the fact that unhappy customers are the likeliest to voice their dissatisfaction, becomes very important for the businesses to provide customer support.

Chatbots Simplified

When was the last time you accessed your email or phone and noticed that your customers have left umpteen number of support requests? Let’s look on the brighter side – if you are lucky, quite a few of these support requests will be fairly simple to process. Now imagine that you had an assistant who could help you process such mundane requests so that you can focus your attention on the strategic initiatives.

While this is an over-simplified example of how a chatbot could potentially help you with your operations, there are a lot of intangibles that need to be addressed before an organization moves towards implementing such a solution.

The first and foremost of which is the “what”, the problem being solved. The organization needs to have a clear understanding of what they are looking to achieve. This could be something as trivial as setting up support call appointments for the company.

Once the what has been, we have to narrow down the tone of the responses identified based on the company’s brand persona. This is important for companies as this helps lend the conversations a personality, if they choose to. A good example of this is “Clippie” the assistant that is widely known to all Microsoft Word users from way back in time. This was Microsoft, as a company, attempting to lend a helpful tone to a chatbot that was targeted as a helpful assistant. The tone is an opportunity for the company to lend it a marketable feature. This can be even via emojis or colloquial verbiage.

At this point, the company needs to map out the value chain on the consumer’s end. This is necessary for the understanding of the steps that a customer would engage in. Such insights can be sought via interviewing the customers or following customer interactions across the brand’s website, for instance. Mapping the value chain of customer interactions is another example of how this can be achieved. This is necessary as in a conversation-based interface, every question needs to be clearly put forth and should be easily understandable for the customer. The questions formed need to be looked at from the lens of customers and the assumption made that there will be misunderstandings. Due to the nature of language and its potential for ambiguity, this step that needs to be paid due importance. The above-mentioned steps are a breakdown of how a company can go about setting up a basic chatbot on their websites.

Challenges we need to take into consideration

While the outlook might be great, we must not forget there are significant challenges that need to be addressed.

  • Natural Language Processing (NLP)
    Do you remember the last time you interacted with Siri, Alexa or the Google Assistant and got frustrated that it did not understand what the query was? This is a common challenge with all such solutions and not just the voice-based ones. The NLP methodologies are still developing and cannot yet understand complex sentences.
  • Emotional Intelligence
    The contextual interfaces in some cases have significant difficulties in understanding the context based on a person’s emotions. Natural language is a complex beast and building an understanding of this beast takes considerable time and effort.
  • Integration
    As more of these interfaces are developed, their integration has left more questions than answers. The deployment of the same across a website and another property has not been easy from the onset, making it draw on additional resources from the company and sometimes requiring duplication of efforts.

We, at Prescience, can help you in taking this to the next level by implementing Artificial Intelligence into the mix of conversation chatbots. One of the salient aspects that we are very proud of in our product offering, is the use of data analytics results towards strengthening our contextual interfaces to help better serve our customers and their end customers. By leveraging these results, we are able to target and provide, with a high degree of accuracy, the kind of questions or queries the consumers might have. An adoption of this approach, we prevent the customers from getting frustrated by the interface and instead, help achieve a satisfactory result in the easiest way possible.

Shivakumar Krishnamurthy

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.