Sterlite Technologies (STL), a global fiber optics provider, has been increasingly using artificial intelligence (AI) to automate the company's employee recruitment and hiring, conduct staff pulse checks, and gamify worker rewards.
The India-based company employs about 8,000 people across the US, China, Europe, India, and the Middle East. Since rolling out its first AI-based applications three or so years ago, Chief Human Resources Officer Anjali Byce has continued to experiment with new uses for the technology, and is currently eyeing how generative AI tools such as ChatGPT could create chatbots to converse with employees for a variety of purposes.
One of the newer uses of AI has been to automate CV stacking and to match skills inventories with specific use cases inside and outside the company.
Along with gamifying rewards, and recognising employee accomplishments that might otherwise be overlooked, STL is now looking at AI apps that can recognise skills deficiencies and recommend training for staff. The automation tools are also being deployed to help workers who are laid off find jobs in other companies — even competitors — through an external network where skill sets are highlighted for would-be employers.
The following are excerpts from a Computerworld interview with Byce:
When did you begin using AI for various HR purposes and why?
AB: "Maybe three years or three-and-a-half years ago. We used it in much smaller ways at that point..., like every employee has a virtual assistant of sorts. You want to apply for leave, you just go into the system and ask it: can you please apply for leave for me for the 29th of June. It will apply for leave for you, send a note to your manager and add it to your calendar, block out time, cancel meetings, or whatever. So we used it in ways to make employee work easier.”
Did you initially build your own AI functionality?
AB: "Yes. We used simple [bots] and algorithms in the backend. That was a simple, fun, and easy-to-work kind of AI application at the time. We then moved onto gamification applications.”
How did you use gamification and how did AI enable that?
AB: "We had a high amount of onboarding that needed to be done virtually. When lockdown happened in COVID, we needed to onboard like 800 people…in multiple batches of 200 people each, things like that. So, we gamified it. We made it a spaceship. You're going to take a tour like astronauts and earn gold coins.
The serious journey in AI started just after that. I would say it’s now part of the entire employee lifecycle — hire to retire. We had a resume-stack ranking process. What basically happens in that is the world is moving from purely role-based jobs to skills-based jobs. Most progressive organisations are already moving to that. So, when we do large-scale hiring, or even if you need to do very sharp, targeted hiring, it would become impossible to scale with speed unless you went out and use AI.
We have an AI-enabled tool that at any point in time, it'll stack rank all the resumes based on the capability of the skill that you may have identified. ...It makes it very smart for recruiters, because if you're doing mass scaling like we do for some of our businesses, you want, say, 200 or 250 resumes of the right capabilities, set and ready — without it being a manual process. It's a very efficient way of really doing that. And that's what the resume- or CV stack-ranking process is. It’s an AI algorithm that runs in the backend.
The other biproduct of this is it mitigates the risk of bias subconsciously creeping into the recruiting process. Because it’s no longer an individual stack or shortlisting of resumes, it’s a structure approach. That’s the other fantastic outcome along with speed.”
Experts say AI can have baked-in biases, because AI is only as good as the developer who created it. Have you seen that?
AB: “It’s not just the stack-ranking process it can mitigate. One of the things we’ve done is all of the job descriptions are gender neutral. So, it doesn’t take in he, she, or etc. Another thing research has taught the industry at large is there are differences in ways men and women project in their resumes. So, there’s a lot more on competence and behavior, etc. But when you stack rank on skills, everybody’s listing skills and then you have a level playing field of skills.
...Does it completely eliminate bias? Of course not, but then you have the next process, which is the screening of interviewers. But you’re right, if you’re not watchful it then depends entirely on the algorithm.”
In what ways is STL increasing its use of AI for HR?
AB: “We started shifting to really looking at AI coming into the onboarding to offboarding process. It was not so much scale and speed we were seeking alone, but it was a better employee experience. We wanted to be sure that there was a consistent experience for employees or candidates joining us. So, we started the journey asking how can we start engaging with employees from minus 30 days from coming onboard, to day zero joining [the organisation], to plus 30 days after day zero. So, it became a very interesting journey using AI in the onboarding process.
Then we did a check with employees who had joined and asked, 'How was your experience?' We got a fantastic 4.8 out of 5 score, or something like that — a really high score in the candidate experience in the onboarding process.”
Can you offer some examples of what’s handled by AI in the onboarding process?
AB: "For example, let’s look at minus 30 days before joining. At different points in time, you want the employee to get access to press releases you're doing, announcements on things happening within the company, some policy documents that they need to get signed up, some confidentiality agreements they need to sign as a part of the preparation for joining, etc.
So, there could be a set of standard things. There could be pulses. There could be newsletters you send out to keep the engagement high — just a whole host of things that you might already configure in the system. It goes in, picks it up and all those who are lined up to join in the minus 30 [group] would get it. It's very personalised. In fact, when the offer goes out, it’s automated. I don't need to do it.
It has my picture with a personalised message from the group CHRO: 'Here's your offer. I'm looking forward to you.' It’s a high tech, but also high touch experience we’re building.”
Did you build out this AI capability yourself or partner with a third-party vendor?
AB: "It’s a mixed bag. So for example, the resume stack-ranking tool is [an] off-the-shelf product, but then you configure it for the way you want it to be. The onboarding and offboarding [tool] is also an off-the-shelf solution, but then there’s a lot of personalisation you have to do, but it’s really worth it from an experience perspective.
The virtual assistant and the other earlier AI tools were all in-house. We used bots and configured them. Our IT colleagues partnered with HR. We told them the need and then they partnered with us to develop it.”
Do you use ChatGPT or the GPT large language model in any way for your HR automation tools?
AB: “We’re not officially using ChatGPT for work, but I know colleagues are having fun with it in different dimensions for different needs. I think what we want to do in the future…, where it’s really going to get interesting is just around the corner for pulse checks. Most organisations do them once in the year — engagement surveys to see how’s the culture.
We’re going to a chat pulse check. It would say things like, ‘Hi, I’m Anjali and I’m a cultural assistant and I want to check in with you to see how things are. So you can really get fast check-ins. You could respond to it saying this team seems to be having a really great time, but here there seems to be some concerns. It’s real time, and that’s something that we’re really moving toward. It’s right around the corner.
And we're getting a lot more gamified. ...Going back to your generative AI question, I think the place that in the mid-term we would most likely use it would be in the learning and development space. We already have some AI that can offer employees prompts, such as have you learned this? This is a need you have identified, etc. But I think that's in future, and I would imagine that’s where generative AI is coming in.”
What advice can you offer on implementing AI with existing backend HR software?
AB: "Change management is a very critical success factor, I think, for any AI enablement and gamification processes. Even simple automation requires good change management. I think one of the things that in my personal experience has worked very well is making sure you have a strong sponsor team that spells out what is required.
So you have functional expertise from HR, but then you partner with strong functional expertise on the IT side. And when those two meet together, I think you get a fantastic solution. You also have to ensure a strong user interface, which is critical for the user experience to be great. And most importantly, you have to have strong integration into the existing systems.
Also having clear cadences where you're able to review progress. And most importantly, [know] why you're doing it. So, for our employee onboarding and offboarding, the why was about how to create a good user experience.”
What speed bumps did you hit along the way to implementing AI in HR applications?
AB: “I think we averted some of the speed bumps. There are so many different tools available in the industry. If we didn't have a good process, it would be analysis to the end of time. We have a good team that had a clear idea of why we’re doing it, and a strong process of evaluating the pros and cons, which would get presented to the sponsorship team. So: have clear sponsors, a good cadence, you stay tight, joined at the hip. Otherwise, yes, it's very easy to run into that trouble.”
You mentioned in an email that AI helped reduce the pain from layoffs. How so?
AB: "There are two or three different things that we do. The first, we have a very strong employee assistance program. And that's really an online process. We've tied up with an external group. They have trained psychologists and counselors. And that’s not just in the case of layoffs. It's for all employees, their families, their kids, every day 24/7. The purpose is to make sure everybody has access to care, has access to counseling and everybody's need is different.
It could be financial counseling, legal counseling, psychological counseling. It doesn't matter. Then, when we’re in an unfortunate situation of a transition of an employee, we definitely focus on making this this available if the individual was to need it. That’s a critical caring approach to have.
We also offer outplacement services related to skills. So, we’ve partnered with agencies that could pick up employees whose skills fit their needs. We also go out and connect with organisations who are hiring. We are very proud of the capabilities we build in our organisation. But if there isn't an opportunity inside the company, I think it's very important for organisations to step up and say, 'I'm willing to offer this to the industry at large, because life happens.'
...So, it's very important to build those partnerships. And that's something that we've been doing very, very actively. and when we do have to transition employees.”
But how is AI helping in the outplacement services?
AB: "We haven't yet put all of this into AI, but [we have] a future path for it. Right now, we've got the [skills] catalog in place. The desire is to put all of this into a tool, and what AI will then do is…look at industry benchmarks, and say, 'Okay, where is industry in terms of needs? What are the highest demands for skills?' And that’s actually very easy to do.
For example, you create a recruiting tool. You know where the maximum hiring is happening. So, you say, oh, the industry is spiking on the following skills. Secondly, we are using AI to input all the skills employees have. ...Where your employees are in competence levels? Do you have more experts or do you have more novices? Do you have some groups of skills that are not [recorded]?
AI and generative AI would actually go into the system, access your learning capabilities and pop up to the employee saying, ‘Hey, you know, your capability on this skill is low. So, if you want to upscale yourself on the following, here's a training course for you.' AI can be used in many different ways once you have the foundation of skill sets very strong.”