AI and Humans Really Need Each Other

Artificial intelligence (AI) is a topic that evokes dread and fear among employees.

Employers are often excited about savings, increased efficiency, and new revenue opportunities. Many employees, especially the older generation, see their jobs disappearing and themselves replaced.

The truth is somewhere between. In a recent panel discussion held during our inaugural HR & Digital Asia Summit, panelists noted that AI is a good thing. It is also far from ready to take over the role of humans in all decisions.

The panel entitled Meet your robot colleague, was moderated by Dilys Boey, partner, ASEAN People Advisory Services Leader, Ernst & Young Advisory. She was joined by Rajesh Narasimhan, director and board member, TVS Motor Company Ltd; Phan Yoke Fei, senior director, Human Resources and Corporate Administration, Gardens by the Bay; Vince Kasten, regional operations - Robotics and AI, Prudential Corp Asia.

We Learn Differently

Where AI and general IT differ lies in the learning process. Prudential’s Kasten noted that AI gets better over time and with more data. “It is one of the great things about AI.”

It is one of the reasons why AI began as a chatbot in many companies. It is now used for employee self-servicing.

Kasten believed that the real strength of AI lies in augmenting human decision making -- not replacing them. “It is more about how [bots] can help business people get outputs. Where the two are coming together intimately is in the operations side. That is where you actually see benefits.”

Yet, Gardens by the Bay’s Phan advised HR leaders not to underestimate how AI and humans interact.

He gave the example of how his organization pioneered driverless vehicles. “But what we did not plan for was the growth of the plants in the garden to affect the sensors. As we continue to meet new challenges, we decided to put two headcounts to operate those two vehicles. They do not drive but will be able to deal with issues.”

Phan noted that HR leaders also need to be ready for employees to interact with AI differently. He did not think there is a single blueprint but rather one that differs by an organization's employee mix.

“For example, we have a chatbot for people to ask about their annual leave. But telephones continue to ring. So they prefer to relate to other human beings.”

Rethinking AI Companionship

Prudential’s Kasten noted that the way forward is to start changing our mindset about AI.

“Think about how Jarvis extends what Tony Stark does. [Stark] would not have been able to do anything without Jarvis’ help. AI lets people get on with higher value work.

Kasten noted that a lot of employees want to contribute to their business through what they do. “But that is not answering emails and or doing the same thing in Excel. A lot of what we do right now gets in the way of our goals. The potential of AI is to help people self actualize what they want to be.”

TVS Motor Company’s Narasimhan agreed. “It is going to be a collaborative effort [between humans and machines]. There is no zero to one, but there will always be fifty shades of grey.”

Gardens by the Bay’s Phan noted that it also boils down to employee mindset. A learning mindset can help employees maximize the advantage of AI.

“Bring some of the inquisitive mindset you had in school to work. People also learn differently. So, HR leaders need to address these individual concerns, especially when you have a diverse culture. AI can help with this.”

Narasimhan called it "learnability quotient." "If you have people who are not receptive learning, then you are always going to be multiple steps behind. AI can actually help humans to take two steps up in the value chain on what they are doing. So, HR leaders need to look for people with the right attitude and are receptive to learning.”

Explaining AI Decisions

While AI is welcomed in augmenting human decisions, it gets tremendous flak for making biased decisions. This makes explainability an important HR topic.

“A human can always come up with an explanation. We all know about cognitive biases; how people make snap decisions; how we use heuristics that may or may not be supported by data. A lot of times when AI makes a decision, it is hard to figure out why they made it,” said Prudential’s Kasten.

Kasten noted that explainability is becoming necessary for regulations. "If a bot makes a decision on a claim in insurance. It is better to explain how it did it [to a regulator]."

Here, data and training methods become essential. "But we really need to get over [the issue about bias]. Every human has biases. You do not have to accept it, but you can measure it and train it out. But do not look for perfect results."

Gardens by the Bay’s Phan noted that it is where appeals and other processes that “need humans to work with the machine” becomes important.

“And who creates this bots? We. We hold the key to the design. As a result, there will always be a degree of bias,” said TVS Motor Company’s Narasimhan.

Instead, he urged companies to understand the context in which AI is being deployed. "Machines tend to make fact-based decisions. That is good, and that is bad. Imagine if you need to explain why you did not deliver results to an HR manager. Here empathy plays a part."