Technology Artificial Intelligence

IBM India's Raghavan on why the true potential of chatbots remains untapped

IBM India's Raghavan on why the true potential of chatbots remains untapped
Sriram Raghavan, Vice President, IBM Research & CTO IBM India/South Asia
2 Jan, 2019

Conversational artificial intelligence (AI) is under-appreciated, considering that chatbots can be deployed for purposes other than just answering frequently asked questions (FAQs), IBM's chief technology officer for India and South Asia, Sriram Raghavan, told TechCircle.

He added that, for chatbots to be meaningful in their evolutionary cycle, they need to be integrated and automated with business process implementation.

Raghavan said that chatbots, in their first phase, have been successful in automating and addressing customers' queries, thereby eliminating large customer response systems for FAQs.

However, the enterprises need to move beyond that by enabling chatbots to take action after customer requests without human intervention. This would probably aid in business transformation, he said, adding that unless enterprises, which currently use chatbots for their customer service, are able to do that, they will lose interest in chatbots over the long term.

"The second piece is connecting these conversations to business systems, making them able to aid in transforming businesses. What is the capability in a conversational system? We want to do implementations of conversational systems," said Raghavan, who also runs the IBM Research Lab in India and IBM Research Center in Singapore. He is also the global leader for blockchain research at IBM.

With conversations with bots -- voice and text -- being understood and appreciated by end-consumers because of the advent of Alexa, Siri and Google Assistant, enterprises should be taking a lead in implementing such solutions that ease customer pain points, Raghavan said.

"It is about using it as a tool for business transformation from an efficiency and customer service perspective, though there is a tendency to see it just as an interesting interface for customers," he said, adding that IBM's Watson Assistant platform helps enterprises to add additional capabilities, bridging the gap between customer requirement and enterprise capabilities.

According to Raghavan, the enterprises in the next phase should be asking whether they can do conversations with employees and suppliers or whether they can gain customer insights from these conversations.

"It is like I am already investing in automation in my company. So, can I connect it to the conversation to provide faster response and run approvals through the conversations? A customer need could be fulfilled through an automation, but it needs intelligence to do that," he said.

But for this to be successful, Raghavan said that companies need a large-scale change in the way the businesses and their various divisions function. "Finally, the business processes also need to be efficient and agile enough to make all this happen. Conversations need to be enabling technology. This is the business process automation and has less to do with chatbots' capabilities," he added.

Raghavan said that once enterprises get comfortable and start seeing value in it, things will gain pace. "When you do a first-cut implementation, it is the starting point and every piece will not fall in place on day one. End-to-end scale needs many pieces and that will help in return on investment (ROI)," he said.

The engineering team at IBM Labs is already doing experiments by automating the business process for its own internal requirements. According to the company, the technologies being developed on a cloud platform enables it to refresh and improve the capabilities on a monthly basis to suit customer requirements.

While all this can happen a lot faster in the developed world, with even the voice-based chatbots in vogue, it will take a while before a large section of Indians end-customers can benefit as there is not enough data and research happening in Indian languages yet. The different accents add to the complexity.

"The voice bots will take a long time as there can be transcription errors. Eventually, the system uses text in the background. So we don’t want to introduce a level of inaccuracy there. It will happen. Emotions and accent also need to be taken care of in voice. Grammar or broken sentences are a huge problem even in native English-speaking countries. In India, the challenge is always to get the bots to work in Indian languages," said Raghavan.