Re-engineering the interview process - the role of AI and Humans

Palantir, a software startup, paid $1.7 million to settle a racial discrimination lawsuit with the Department of Labor’s Office of Federal Contract Compliance Programs (OFCCP). According to the suit, from a pool of more than 1,160 qualified software engineer applicants where approximately 85% were Asian, the company hired 14 non-Asian employees and only 11 Asian people. Despite a majority of qualified intern applicants being Asian, Palantir hired four times as many non-Asians as Asians.

Hiring Bias has been talked about for several years. Humans are hard-wired to make quick decisions and decision making is often driven by unconscious biases. In an ideal world, the decision to hire a candidate would be based solely on their ability to do the job well. The hire would be approached in an objective, pragmatic way, free from subjectivity and unconscious bias. Keeping this approach in mind, AI was introduced to the hiring process. In recent years, many software firms have come up to sell companies AI recruitment tools. These companies are built on the core philosophy - human recruiters are hopelessly biased while machines are objective. 

But is the solution as simple as it seems? Recently, Amazon was in the news when it developed an AI resume scanner that was very publicly revealed to be biased against women in 2018. Designed to be a benchmark in recruitment AI, it filtered out resumes that contained the word “women” (eg. women’s sport or club) and developed a habit of recommending candidates with no relevant skills or experience.

What went wrong here? The answer is simple - our philosophy is flawed. We tend to overlook the fact that AI algorithms cannot be inherently objective as they are designed by humans. In fact, hiring software can introduce new layers of bias and discrimination and can complicate the process even further. Nonetheless, if designed carefully, AI can still benefit us and solve this problem.

The Role of AI

Let us look at the possible solutions that AI can introduce to us:

‍1. Becoming a scrutinizer 

The power of data and analytics is still unexplored in the recruitment process. Typically, recruiters are so hard pressed on time to fill positions, they hardly spend time to understand how they can hire faster the next time. AI can help us solve this problem by identifying our biases. By understanding these patterns, recruiters can course correct and focus on attributes that are more relevant to the job. 

2. Ranking candidates

Instead of filtering out candidates, AI can be trained to rank candidates. There are many tools in the market which use AI predictions to screen resumes and guide recruiters on who should be contacted first. This is done by ranking skills and criteria that are more relevant to the job. While designing these algorithms, we have to be careful about criteria that we are giving more importance to in a job. 

‍3. Assessing and interviewing

Hiring for a middle to senior level position requires several rounds of screening. This is very time consuming and AI can come to a recruiter’s rescue. A lot of innovation has taken place to introduce automated assessments and personality tests to screen candidates. 

4. Handling Bulk Data

A human needs at least 2 minutes to screen a CV while AI only needs 3 seconds ! For blue-collar jobs and bulk hiring, it is helpful to have AI perform most of the tasks that would typically be taken care of by humans.

Where can humans improve ?

While AI can help with automating manual tasks and providing data insights, humans are still important for all the decision-making processes. The problem of biased decision-making by humans is valid but it can be solved if humans follow a pragmatic approach in the interview process.

1. Standardisation

Interviews are mostly unstructured in most organisations. Standarsised and structured interviews can lead to higher efficiency. Each question can be designed to assess a specific skill or attribute which has a direct linkage to the job. If an interviewer is prepared on what he/she needs to ask, there is limited scope for bias. Additionally, it leads to a better candidate experience since all candidates are screened and evaluated on the same criteria. 

‍2. Interview Rating Scale

Once a structure is in place, interviewers must establish interview rating scales. In an interview rating sheet, the interviewer gives a candidate a score based on how well they answer a question. All scores add up to give a combined score which can then be compared with other candidates. This data can not only help in hiring the right candidate, but can also be used by AI to predict other factors and improve the hiring process.

‍3. Panel Interviews

Panel interviews conducted by a heterogeneous group of interviewers allow a balanced and unbiased selection. Collectively, a group can follow a logical approach towards assessing candidates on their accomplishments without being driven by individual biases. 

The future of hiring sees both humans and AI working as a team to improve turnaround time and hire the right talent for organisations. Humans would be playing a role of overseeing the function and taking crucial decisions but the role of AI would also become indispensable. Overall, we are going to witness the transformation of the hiring function from being manual, redundant and opaque to becoming data driven, automated and efficient.

‍