Person holding a mobile phone on a wooden table engaging with a chatbot.
Image: © WrightStudio/

Could your next first-round interview be with a chatbot?

17 Sep 2019

Facing a deluge of applications, some of the world’s largest organisations are increasingly turning to AI and chatbots to help in the recruitment process.

Future of Work Week graphic.

While the technology used when applying for a job has changed dramatically in the internet age, the interview process is still largely as familiar an experience as it would be to someone a few decades ago.

For recruiters, this means sifting through reams of email print-outs and trying to come up with the easiest way to get through thousands of applications. In addition to the strain this would put on any human having to do such a task, there’s also the issue of bias that too commonly crops up in the hiring process.

This issue has led some to suggest that artificial intelligence (AI) and machine learning can at least be the first port of call to make a recruiter’s life a bit easier. During a recent visit to Singapore, I caught up with one start-up that has already caught the attention of multinational organisations with the promise of “making hiring fairer”.

Headquartered in Singapore – but with offices in India and the US – Impress is a start-up co-founded and led by Sudhanshu Ahuja, which has developed a recruiter chatbot designed for the preliminary stages of the interview process.

When an employer selects a list of competencies required for a job, the chatbot is published along with the job posting and can then respond to applicants. If a candidate meets all the criteria, then the recruiter can start reaching out to them for the next round.

So far, the concept seems to have ticked some boxes for a number of organisations, with Impress’s chatbot involved in hiring in its native Singapore with the banking giant DBS and Singtel, but also internationally with Accenture and Panasonic.

DBS has reported results for the JIM (Jobs Intelligence Maestro) chatbot, claiming an 81pc reduction in candidate qualification time with 400 people recruited annually using the tool. Last September, Impress raised $1.2m and plans to raise more this year.

Working with the Singapore government makes up half of the company’s revenues, according to Ahuja, but he is now using this as a basis for expansion into work with the US government, or at least as part of preliminary discussions.

The bias problem

But regarding the technology, there is an obvious question that you might be wondering: could a biased AI chatbot unfairly eliminate a human candidate just as much – or more than – a human recruiter?

Bias in AI is not a new concept, with many researchers pointing out various flaws in algorithms when it comes to making important life decisions, or especially when related to facial recognition.

But Ahuja said that he and the rest of the Impress team developed the technology fully aware of the perils of ‘black box AI’ – an artificial intelligence system where decisions cannot be explained.

“We started from the mindset of making hiring fairer. We know the bad things that can happen from the wrong uses of technology, we stayed away from deep neural nets,” Ahuja said.

“Our tech isn’t a black box, it breaks it down into competencies. The machine learning happens within these competencies – you break down the black box problem into smaller pieces and then you make sure you’re following the best practices in each of them.”

AI interviewing AI?

In fact, Impress took a stance that almost all of the content in a person’s CV would remain outside the hiring process.

“We don’t rate the CVs at all,” Ahuja said. “That’s where most of the bias comes from. Even when companies ask us to do that, we say: ‘Sorry but that’s a dangerous territory and that goes against certain principles’.”

Experience written in the CV is anonymised, he added, with approximately 80pc of the decision to hire the candidate based on the answers given to the chatbot. While the Impress team seem to be happy with where the technology is at right now, Ahuja said that the future of HR chatbots lies beyond just preliminary interviews, particularly during the onboarding process.

He described the combination of competency for a role and a person’s performance as a “holy grail”, providing employers with decisions that deliver the highest financial returns.

In that same vein, I asked whether – somewhat ironically – does Impress fear the rise of automation with an increasingly AI workforce threatening its business, or if AI will inevitably interview AI?

“We have thought about that,” he said. “We think that’s about five or more years away. But you’re going to have companies that are going to be looking for AI-plus-human combinations in the future for many jobs.

“In fact, even if the economy goes bad, businesses need to decrease cost, which is what we do. Both ways the scenarios are positive for us, we just probably have to adjust according to where the market is.”

Disclosure: The journalist’s trip to Singapore was provided by the Infocomm Media Development Authority

Colm Gorey
By Colm Gorey

Colm Gorey was a senior journalist with Silicon Republic. He joined in January 2014 and covered AI, IoT, science and anything that will get us to Mars quicker. When not trying to get his hands on the latest gaming release, he can be found lost in a sea of Wikipedia articles on obscure historic battles and countries that don't exist any more, or watching classic Simpsons episodes far too many times to count.

Loading now, one moment please! Loading