In a break away from tradition, we speak to a significant HR stakeholder in any change management effort – Aditi Sharma Kalra learns HR-relevant guidance from the experience of Ravi Madavaram, Head of Artificial Intelligence, Axiata.
Q What does artificial intelligence mean to a swiftly transforming telecom major?AI is being used as a buzzword in the market, so it’s essential we define what we mean by it. We categorise three main types of algorithms as AI – supervised machine learning, unsupervised machine learning, and reinforcement learning. There are various use-cases where each of these algorithms have application.
My primary responsibility is to identify the business problems that can be solved using AI, then work with various teams to develop a proof of concept (PoC). If the PoC is scalable then we convert the PoC into a fully fledged product.
Q How closely do you work with the head of HR, and on which kind of projects?Specifically with HR, we have worked on employee communication and employee query resolution use cases. We also have automated IT help desk queries. We are working on employee satisfaction and churn prediction, as well as the hiring process automation.
Q With many companies looking to digitalise parts of their operations, what does it take to build in-house expertise in tech?Having an in-house tech team is not always advisable. It made sense for us since:
- We have 10 operating companies with over 10,000 employees.
- It was impractical for each company to hire their own expert – as the utilisation of the expert was <50%.
- We could reuse a lot of solutions that we developed for one company in another company.
Also, the in-house tech team needs to identify which part of the skills are replicable across various use-cases and only hire full-time employees for this. The rest of the skills need to be complemented through contract employees and vendors.
Q What’s your take on the way forward for AI, especially for people processes?Over the past three years, a key area where the adoption of AI stalls is the human-machine interface. The AI that we build is never 100% accurate and when it fails, we need to plan and deploy a streamlined and smooth transition for a human to take over. This is challenging in many ways, such as in the responsiveness of humans and AI’s ability to identify the process breakdown. The more legacy systems a company has, the more challenging this becomes.
The other challenging part of AI is the testing of the solution before deployment. Since AI is built for an open-ended input, the response of AI varies significantly, unlike traditional software solutions. This puts a lot of stress on the people involved.
Photo / Provided