Replicating human intelligence in a machine has been the holy grail of computing ever since the first computer program was written, but our expectation of chatbots is sky-high

“Alexa, order some flowers for my Mom.”

“Anita, what’s the next best stock to invest in?”

Each day, millions of people around the world are asking such questions to, well, chatbots.

From ordering flowers using Amazon’s Alexa to receiving advice on your next financial investment, chatbots are slowly creeping into and becoming a part of our daily lives. Whether we like it or not, these “intelligent machines” that can converse in our language and can display unparalleled flawless intelligence have a rosy prospect, more so when you consider the market trends.

Understanding and replicating human intelligence in a machine has been the holy grail of computing ever since the first computer program was written. Hollywood’s depiction of AI has further added fuel to this vision. Needless to say the expectation we have from our “chatbot” friends is sky-high.

It might, therefore, be time to take a step back and look at some of the common myths about bots doing rounds in the market.

But first, what is a ‘chatbot’ exactly?

A ‘chatbot’ has a conversational chat interface (similar to a WhatsApp chat box) and has an intelligent brain behind it that responds to the user. Typically, these chatbots are deployed for specific purposes such as sales, service, education or sometimes, for pure fun.

MYTH 1: Artificial Intelligence (AI) is incorporated in all bots

Well, not really. It turns out that this is a common misconception that all chatbots are driven by AI technology and can think and understand customer intent accurately.

There are 3 categories of chatbots in the market today:

Script-driven bots: These rely on pre-defined scripts to answer queries, following what can be viewed as a large flowchart. This category of bots, which includes some of the well-advertised ones, are essentially not using the AI technology stack. Their advantage is that they are easy to setup but notoriously difficult to maintain.
Data-driven bots: This class of bots is good in recognizing patterns based on their previous learnings and responds from their catalogued responses. They may use AI for pattern recognition, but have no clue of what is being discussed. These bots are like smart parrots, which greet you when you say a certain word that you have taught it. Again, the necessity of a large database is their limitation.
Inference-driven bots: These use cognitive models, natural language processing and deep learning methods to infer customer intents. Such bots have true AI capabilities to build knowledge maps, learn from previous conversations, contextualize and respond like a human would.

MYTH 2: Chatbots are easy to develop

A simple chatbot is easy to develop within a few hours using some of the commonly available APIs. However, chatbots that simulate human cognitive capabilities are not as easy to develop and require robust technical knowledge in AI.

Companies that have domain expertise and AI skills are able to deploy complex solutions in a short span of 6-8 weeks. A suitable example is the chatbot developed for ICICI Insurance, “MyRA”. MyRA is responsible for end-to-end management of the customer journey from helping the customer select appropriate insurance coverage, answering pertinent queries, generating the policy and resolving issues that may crop up midway.

Also read: 9 principles for using chatbots, messaging and speech-based assistants

MYTH 3: Chatbots have a human feel to them

They don’t always need to have a personality. However, chatting with a bot that has one encourages us to have a conversation and derive greater satisfaction from the experience. Building personality into a chatbot is a challenge that is not easily achievable.

The technical architect of ICICI MyRA spent years researching and developing technology that serves this need. As a result, the bot is able to carry-out a conversation with a human touch and can even recognize “anger” or “frustration”. It also understands when a conversation is not going in the right direction and seamlessly connects to a customer service representative, ensuring an overall high quality of service.

MYTH 4: AI will solve all problems

Artificial Intelligence, as depicted in Hollywood, seems to give an illusion that all problems can be solved by deploying an AI system. However, this is far from the reality.

To begin with, not all use cases are ideal for an AI system. Systems that need 100% pre-programmed logic or where the cost of developing an AI system outweighs the return on investment are not appropriate for automation using AI and chatbot. Secondly, for systems that do implement AI, if the scope of the application is not controlled from the start, the project is destined for failure.

MYTH 5: Chatbots will replace humans

Chatbots will never replace humans. Rather, they augment and optimize human potential in a smart way. When you have the ability to help customers with intelligent, instant and to-the-point answers for their questions with a great consistency (and also while keeping their preferences in mind), chatbots do emerge as a more favourable option as an initial customer touchpoint. This also minimizes the workload of technical experts, who otherwise have to address recurring non-technical queries, thus, considerably boosting the overall efficiency.

Conclusion

Artificial Intelligence has captured the fascination of mankind since the very first thought about intelligent machines cropped up in the human brain. And why wouldn’t it? It is one of the most phenomenal developments that has the potential to truly redefine the future. Since such systems are programmable and build upon the edifice that we provide them, they have a pivotal role to play in making the ultra-modern world, which we so often envision for our society, a reality.

—-

Editor’s note: e27 publishes relevant guest contributions from the community. Share your honest opinions and expert knowledge by submitting your content here.

Featured Image Copyright: galitskaya / 123RF Stock Photo

The post Here are 5 myths about chatbots that people are getting completely wrong appeared first on e27.