AI solutions to help pre-empt and self-heal operational issues

The challenge with RPA is that it interacts with an ever-changing environment which becomes an overhead to maintain and needs to be constantly upgraded with rules.

In 1950, Alan Turing proposed that machines can do what humans can do. To prove it, he developed the Turing Test to demonstrate intelligent behavior by machines, indistinguishable to humans. Turing proposed that a human evaluator would judge natural language conversations between a human and a machine designed to generate human-like responses. If the evaluator cannot reliably differentiate machines from humans, the machine is said to have passed the test.

Since then, machine learning and artificial intelligence has come a long way. Today, there is a much higher level where artificial intelligence in the telco domain makes it ever hard to tell whether it is the machine or the human responding to any customer engagement processes, managing operations or tuning the networks.

The systems today are buttressed with the computing power that was not available before and therefore, we are privileged with the opportunity to deploy learning algorithms that can build self-learning systems supported by big data and the internet of things (IoT) that is capable to digitize the surrounding environment, providing the much-needed feed to the learning algorithms.

Equipped with such machine learning capabilities, the limitation is then to
imagine what can be done with systems that have such artificial intelligence. One such example in moderation of agriculture using IoT shows the importance of such imaginations.

In a classic environment, the operations engineer is engaged to attend an issue when the alert is raised. But the AI systems can establish patterns and predict the impending problems much in advance.

So, before the issue is manifested, the cause is fixed, and balance is restored.

While this one case might be easy to imagine, the benefits from AI
today are limited because we are still in exploration phase on the possibilities where such solutions can be applied. In the similar direction, a powerful example in medical industry demonstrates how AI is used to halt cancer progression.

Think of scenarios like network faults which can not only be costly and time consuming to repair but can cause huge disruption to the business impacting multiple subscribers in a short span of time.

While waiting for alerts and monitoring by humans on the dashboard can lead to a fast solution, the challenge today is different. It is to smartly
capture all the data points much in advance and intelligently correlate to understand what kind of problem can arise.

Such a system will also learn solutions for such problems and applies them on a real time basis. Humans and scripts are no match for this kind of
speed to process and apply fixes. It changes the system to become autonomous instead of being automated to run certain jobs.

Another example could be sentiment analysis. There are highly advanced algorithms that can tell the tone of the conversation, written or spoken, and understand the sentiment of the consumer.

For a human it might be impossible to find out what is causing the stress to consumer. Not even by running several scripts and manually correlating the results for the same. But an AI system can efficiently collect the data
from several actions and interactions in the recent past and beyond to understand the impacted parameters that would lead to customer dissatisfaction.

It can also do the course correction and suggest what is the best approach to make the customer happy, due to its learnings from various customers which again would have at best, limited understanding with any given human representative based on recent memory of interactions.

This is making things possible that looked impossible hitherto. Consider an
example where Microsoft glass helps the blind to see! Even several humans could not have helped the person the way AI helps in this case.

There are several more examples which are in different stages of evolution and again, are only limited by the imagination that will determine the maturity of operations in the future.

Operations will not remain a boring back-office job for some operators, but
a living interaction of technology with people. It will take away the monotony of the mundane tasks, and provide people with advance information to take intelligent decisions.

It will be possible to bring the data from multiple systems in the hugely
interactive ecosystem and draw several correlations, and learn over a period what matters. Operations will soon be the biggest playground to make things happen with AI and make a difference to the lives of the people.

(The author is Head of Engineering, Amdocs Intelligent Operations)