May 7, 2021

Knews

Your News Buddy

Machine learning: Everything HR technologists need to ask vendors

Angela Ashenden, principal analyst, workplace transformation at CCS Insight, told me that machine learning was increasingly applied in HR settings to help scale and automate resource-intensive or repetitive tasks:

“We know that one of the most common areas of deployment is in recruitment, helping to take some of the heavy workloads off recruiters by using AI to screen resumes, particularly in the early stages of recruitment.” 

Ashenden further noted that personalized training recommendations for employees are another common machine learning application in the HR space.

Personal training recommendations can help companies leverage existing information about the employee experience. Employers and HR experts might also gain insight into other career aspirations — allowing them to extend more direct involvement in employee development paths. 

Bots, sentiment analysis, and more 

Monitoring the workplace market is a full-time job for Ashenden who told me she noticed a growing interest in AI-enabled chatbots, which often get deployed on mobile apps, company intranets, or collaboration platforms such as Microsoft Teams and Slack.

The deployment of natural language processing (NLP) means employees can get quick responses to questions around company policies. The technology also enables HR teams, which have traditionally been swamped with admin-heavy tasks, to automate processes such as requesting time off or submitting expense reports.

Sentiment analysis is another area where the technology is showing promise. With much of the world’s labor force working remotely, it’s never been so important for organizations to get a sense of how employees are feeling.

Luckily, the market is awash with vendors addressing, some of which lean on machine learning, to address this very issue.

Examples include LinkedIn’s Glint, Peakon — which was recently acquired by Workday — Qualtrics, and Medallia.

The challenges: deploying machine learning in HR

As with many technologies, deploying machine learning isn’t always easy and in HR’s case requires significant considerations about internal learning, and investment.

Amin Venjara, chief product owner and vice president of product, DataCloud, at ADP, a people analytics, machine learning, and reporting platform, agrees. There are difficulties surrounding people and skillsets because deploying machine learning requires a more advanced skill set than traditional HR departments often have, Venjara adds.

He shared that if HR departments decide to build out their people analytics and machine learning capabilities, they first need to form a more technical team. That team should consist of data engineers, scientists, and visualization experts, a small feat for larger organizations and a potential setback for smaller, more vertical-focused industries.

Other issues, including the depth and quality of data, can impact a company’s desired outcome in most cases. These models, according to Venjara, must be well-trained as they can influence the accuracy of results.

Artificial intelligence often gets touted as a great solution to many of the world’s problems but there are, however, some concerns surrounding ethics and bias in trained AI models.

These models are more often than not, built and trained by an industry that is overwhelmingly white and male. And this can present an unfair disadvantage in the HR process, notably, if data results in the product of human decisions – who gets hired, who gets a raise, and who gets promoted. 

Driving adoption also remains a pain point for HR professionals when it comes to deploying machine learning. HR technologists have to get buying from budget holders, colleagues, and more importantly, employees.

In order to unlock the true potential of machine learning, HR technologists should make sure the solutions they invest in and role out are user friendly.

what HR needs to ask vendors

Solving many of the issues discussed thus far is not impossible, but it will take time — and industry vendors seem to be working toward remedying some concerns plaguing businesses. 

Some of the responsibility lies on the end-user, though, and HR pro and technologists should inquire about the depth and quality of data used, meaning that data should be contemporary, relevant, and accurate for training purposes. The maturity of the model you are using impacts the quality of candidates your organization receives, so asking vendors the right questions is warranted. 

I’m told this includes questions like how often are models re-trained, and how are models monitored. By doing so, customers (aka HR leaders) should be able to gain better insights about how to fix issues, along with what the potential impact might be. Driving adoption entails focusing on an often underinvested component of machine learning — the impact of its user interface. 

“By this, I mean; how easy it is for a consumer of a machine learning model to understand why the model is making the classification or prediction,” Venjara concluded. The result, eventually, is to develop a platform that is so straightforward to deploy — it poses almost no learning curve, and it massively drives user adoption. 

One of the key takeaways from my conversation with Venjara is that we must remember the goal of these kinds of systems, not to replace, but rather to aid the HR professional in the decision-making process when choosing candidates, relevancy, and risk.

Real-life context

I also reached out to Josh Brenner, CEO of Vettery/Hired. Working with the likes of CapitalOne, ShutterStock, Zendesk, Humana, and iHeart Radio, Brenner added that the benefits of leveraging machine learning technology in the HR process are unmatched, noting that: 

“From a recruiting perspective, hiring managers at large companies often need to sort through hundreds or even thousands of applicants for a given position.”

Leveraging an advanced AI-based platform to gain insights can cut out “significant time” spent on pre-screen interviews, inbound resume screens of unqualified candidates, and hours spent messaging unresponsive passive candidates. 

“Our data show that you can improve many processes as much as four times — which you can instead spend meeting with and assessing diverse talent.”

Technical sourcing lead at Dropbox, Kenny Koran, said the company deployed HR technology that uses machine learning, leading to a constant stream of more qualified vetted candidates. 

“We were able to see who we made an interview request to, who accepted our requests, and how we are doing compared to others on the market. In a competitive market, that helps,” he adds. 

Technical recruiter at Peloton, Kevin Minchella, said the company’s greatest pain point is keeping up with candidates and keeping them happy. AI-based tech helped the organization vet more skilled candidates it says were “interested and actively job seeking.” 

I’m sure you are familiar with Capital One. It is one of the largest financial institutions in the world. Hayley Metcalf is a senior tech recruiter at the firm, and she told me, after four years of using machine learning-based HR software, Capital One hired over 200 candidates. 

These individuals often fill what she calls vital positions. Capital One is also said to have increased its interview acceptance rate to 60%. All this could have lasting impacts on the digital transformation strategy of most, if not all, companies who take advantage of the technology and deploy it in a frictionless way.