AI and ML allows Efficient Data Organization and Data Quality Management: Shanmugam Nagarajan, [24]7.ai

Pradeep Chakraborty Jun 25, 2018

[24]7.ai is redefining the way companies interact with consumers. It helps businesses attract and retain customers, and make it possible to create a personalized, predictive and effortless customer experience.

Shanmugam Nagarajan, Co-founder and Chief People Officer, [24]7.ai, tells us more. Excerpts from an interview:

BW CIO: How will AI applications drive productivity and business performance?

Shanmugam Nagarajan: For us at [24]7.ai, artificial intelligence does not operate in a silo. It is the ultimate combination of artificial intelligence and human intelligence that we are excited about – and the intersection of the two spurs the productivity and performance gains.

For example, in the realm of customer interactions, AI is already helping us immensely by helping predict customer ‘intent’ – what is it that a customer is trying to achieve, which, in turn, cuts on the time to service and vastly improves customer satisfaction. The second aspect is blending chatbots and human agents to deliver a delightful experience.

If done well – that is, if the bots are able to assist customers accurately, speedily and without losing the context, then a lot of repetitive transactions can be contained by the bots. Human intervention takes place if the bot is not able to address a query and the full context of the conversation is passed on to them. This approach is a huge leap in improving the efficiency of customer service operations.

We envision a future where there is seamless back and forth between the agent and the bot depending upon the nature of the queries. Some parts of the transaction can be addressed by bots and the others by human agents. If you were to look at AI’s applications in different industries, there are several examples of how it can spur productivity gains.

In the retail sector, AI can help tremendously by offering personalization – empowering enterprises to make the right offer at the right time to the right person and also facilitate optimal inventory management through AI-enhanced forecasting.

The industrial sector enjoys several use cases already and it will witness further efficiency improvements. AI-based visual inspection for improved defect detection, a combination of AI+IoT for predictive maintenance, smarter power grids, AI-enabled autonomous cars; these are just a few examples of how AI can positively transform the industrial sector.

In the healthcare sector, AI can augment human capabilities by managing medical data, compiling and analyzing information from x-rays and scans, developing personalized treatment designs based on patient history and available medical knowledge and so on. Based on the analysis of the historic data of success rates of various treatment options for a particular ailment, doctors can choose the treatment plan with the highest probability of success.

These kinds of AI interventions could save a lot of time for medical professionals by augmenting decision-maligning and help them treat more patients more successfully.

BW CIO: How can AI help in processing and analysis of data?

Shanmugam Nagarajan: There are several ways how artificial intelligence and machine learning (ML) can help in processing and analyzing massive amounts of data. For the companies that deal with large volumes of data, everything related to data processing has to be done at scale, and AI can help in that.

Specifically, AI and ML can be used, for more efficient data organization and data quality management, and therefore, improved data processing. Some of these areas include intelligent detection of data schemas, automatic data quality control, detection and fixing of anomalies in data as well as intelligent indexing and storage.

The last one is an emerging field where traditional indexing technologies for large-scale datasets are being replaced by ML-based indexing methods.

In terms of analysis of large volumes of data using AI and ML, several use cases exist. Using ML models, specifically natural language processing models, one can extract the important information (profile data) about any entity from the unstructured texts.

The other two emerging areas include detecting security breach/data tampering in large data storage, and detecting anomalies in processes that contain multiple steps. Using AI and ML works especially well for the latter, where each step of a process produces data which is consumed in subsequent steps.

Last, but not the least – one of the most popular applications of AI and ML models in data analysis is in monitoring. AI enhanced monitoring can be used in multiple scenarios, for example, compliance monitoring for monitoring systems and processes to ensure compliance with the existing laws.

Another use case would be using AI and ML models in process details monitoring for semi-real-time reporting, if the data processing speed is not matching up to the data volume. AI and ML models can also be leveraged for monitoring conversations in real-time to understand emerging topics, for example, on Twitter or Facebook or Google search.

BW CIO: How much is there an increasing need for robust data storage solutions with growing AI applications?

Shanmugam Nagarajan: For AI to predict better and be intelligent, it needs varied and massive amounts of training data. As AI applications mature, newer data streams will be added to further enhance intelligence and improve prediction, thus, continuously increasing the overall volume of data that needs to be stored.

I came across this example of the Autonomous Automobile Industry study that shows that every week about 8 to 10 Petabytes of storage is consumed. Apart from collecting data from automobiles, other data streams from various other sources like Weather, GPS, etc. are added. As more and more features are introduced, new streams of data become available and are incorporated for enhancing the AI capabilities.

Given this changing context, the traditional storage solutions will find it extremely challenging to predict and provision storage on the fly, in a timely manner for such a scale and at a reasonable cost. Hence, many companies that use AI as an integral part of their business, have started looking for newer methods such as the Scale-out-based Storage Architecture.

The Scale-out-based Storage solution provides horizontal scaling, along with computation requirements, taking care of both scale and speed in parallel.

BW CIO: Do companies have clear strategy and direction? What tips would you give them

Shanmugam Nagarajan: There is keen awareness among business leaders that AI has the potential of fundamentally changing the operations of varied industries ranging from agriculture to aeronautics, and from travel to telecom.

We see a great sense of urgency amongst CTOs and CIOs to ensure their organizations don’t lag behind in the new world order that is being shaped by AI.

Like any other new innovation, AI adoption will also follow the technology adoption curve – the innovators and early adopters will embrace AI applications rapidly and with great enthusiasm.

On the other end of the spectrum, the late majority and the laggards will wait and watch for AI to fully mature before they give it a go.

However, given the buzz around AI, it may seem a little overwhelming as to what your approach to AI adoption should be. A good roadmap would be to understand the potential of AI for your industry, prioritize use cases based on RoI, and have strong business cases that justify the use of AI in specific scenarios.

Beyond deciding where and how to best employ AI, a culture that celebrates the combination of artificial intelligence with human intelligence and creates a conducive environment for this approach is crucial for any organization to reap full benefits of AI.

It’s also important to ensure that you have the right AI talent available. In a recent BCG survey almost 70 percent of executives surveyed acknowledged that their company lacks the AI skills to speed up the introduction of new technologies. Therefore, organizations must focus on upskilling their own employees or acquire AI talent from outside of the organization to ensure success.

These are still early days and the field is still evolving. One must also be ready for a little bit of experimentation as all deployments may not yield desired levels of results immediately. In such cases, you must be quick to analyze the impact, introduce tweaks and make objective assessments in terms of the business viability of such deployments.

BW CIO: What is the scope of AI in India?

Shanmugam Nagarajan: Our Prime Minster has put a spotlight on AI indicating that AI is indeed a priority for the government. Directionally, the Indian government has taken the right stance by setting up the AI Taskforce and is also actively pursuing public-private partnership approach.

The recent discussion paper floated by the Niti Ayog on national¬ strategy on AI outlines five core areas: Healthcare, Agriculture, Education, Smart Cities and Infrastructure, Smart Mobility and Transportation. The paper has recommendations to build a truly transformative approach in pursuit of #AIforAll and has invited recommendations from all the stakeholders.

These are all the right steps. However, to really become a leader in AI innovation, we need to focus our energies on all aspects of building a self-sustaining AI ecosystem and take steps rapidly, or we may just miss the bus.

In addition to a significant budget allocation and creating a robust public-private ecosystem, other areas that need immediate attention and action include revamping academic curricula, attracting AI talent -- both, in research and industry, and encouraging AI-based start-ups. A well-rounded approach that nurtures the entire AI ecosystem is the need of the hour.