AI/ML complemented by cloud makes automation faster and easier

A blog post by David Linthicum, managing director, chief cloud strategy officer, Deloitte Consulting LLP; and Brijesh Singh, managing director, AI and Cloud Strategy and Conversational Systems

The ability for AI to improve customer service continues to gain traction in the market. In fact, automating call-center interactions by replacing human-to-human interaction with human-to-machine interaction has the potential to establish better customer relations and increased trustworthiness. One reason is the evolution in the domain of conversational AI, which has made the robust conversion of speech to text a reality. Coupled with cloud, organizations can leverage these capabilities on a scale that was not necessarily possible in the past.

What changed in the world of AI

There are three primary entities with which any organization engages: customers, business partners, and employees. The traditional way of engaging these three groups was either via a call center or by sending emails. Then came the era of chatbots, where the voices that respond sound more humanlike, with the enhancements of speech-to-text and text-to-speech technology, also known as conversational AI. This transformed business engagement from occurring between two humans to interaction between a human and a machine. Those interactions that still take place between humans are augmented by these new technologies, including support from various AI techniques to automate repetitive tasks.

We seem to be moving toward a time where human-to-human interaction is going to be reduced to just certain circumstances where the chatbots are not optimal for dealing with unique requests or inquiries. As machine learning powers an engine in the background, it enables conversational systems to disambiguate these calls and create a knowledge graph. The graph plots the predictability of potential answers and can lead to higher call containment (a situation when a caller’s purpose has been solved) and higher customer satisfaction with greater accuracy and assurance.

Cloud as an enabler of AI applications

All these tasks can all be executed faster and more easily than ever before because of the elastic nature of the cloud, its strong compute capability, and the availability of storage resources. Data is now easily accessible by data scientists, who can access platforms where exploratory data analysis can be done easily and leverage prebuilt models available through open-source platforms. Proofs of concept can now be quickly executed, along with demonstrating the capabilities of models, before employing them in a production environment. This entire cycle, which is now known as MLOps, is enabled by the cloud environment. Technologists believe that the power of cloud will lead to a much broader adoption of conversational AI soon. However, it is imperative to ensure that integration of data with other systems is secure and that compliance and regulatory requirements are followed stringently. Success with conversational AI will hinge upon a combination of factors from evolving technology, improved data training, and ensuring a strong governance process.