How to use data and artificial intelligence for diversity and inclusion in hiring and onboarding

Among all the myths about work that COVID-19 has shattered, the biggest has been that employees will keep working at places that don’t value a good culture as long as they are paid well. Many organisations have seen mass exits recently because their culture was not built to bring teams together in the face of an event like this. This is despite early evidence, including from 2019, when 50% of employees surveyed by Glassdoor across all countries believed their company should do more to increase diversity. The number is 76% in the same survey conducted post-pandemic.

Lack of diversity in organisational culture in 2021: intentional or a miss?

It is a safe assumption that there are only a few employers who still don’t see the correlation between motivated and diverse teams and increased productivity. 83% of executives agree that a diverse workforce taps into a diverse client base and innovation, and organisations with diverse boards have 95% higher returns on equity (a Harver Report). Why then is this awareness not transforming into effective action? There are a few reasons:

  • Inexperienced leadership for a cohesive diversity and inclusion (D&I) strategy
  • Insufficient data and tech infrastructure to measure impact of D&I strategies
  • Lack of faith in technologies like AI to enable diverse teams

Making data work for you

The popular adage you can’t manage what you don’t measure is best applied to D&I strategies. Collecting and harnessing data is the first step towards diagnosing inherent biases that prevail in organisations. Each step in the employee lifecycle, from sourcing, verification, selection, and compensation, to engagement, pay, promotion and attrition holds data points that reveal an organisation’s cultural core.

From our experience of handling background verification and onboarding journeys for organisations of different scales, we have a deep understanding of what makes a truly diverse team. The data cues are everywhere — interviewing individuals from diverse groups versus those who are not from a diverse group, percentage of hires from diverse groups versus industry standards, and financial incentives given to diverse employee groups against those who are not from a diverse group — are a few places to look at. The challenge, however, is identifying what diversity and inclusion mean for your organisation and building a responsible and inclusive AI that adheres to set standards.

Channelling the power of artificial intelligence for diversity and inclusion

Age, physical abilities, race, ethnicity, gender, and sexual orientation are primary markers of diversity at most organisations. However, there are secondary markers as well that manifest in the form of education, class, language, geography, marital and parental status etc. To make matters more complicated, organisational bias is often found at the intersection of these. A D&I strategy, in the absence of an informed AI accounting for these intersections, can simply not be effective in championing for the diverse pool of employees that modern organisations look to work with.

What then stops many well-meaning organisations from relying on artificial intelligence for an inclusive culture is a very real threat that artificial intelligence, if unchecked, can overtime exacerbate human bias and create more problems than it solves. Therefore, a combination of human + artificial intelligence has emerged as the best way forward. AI-powered systems can be modelled after an organisation’s inclusive values to reach out to a more diverse pool of candidates, scan their resumes faster, verify them against relevant parameters of their diversity marker, and onboard them within a cultural context that they best identify with.

Assimilating convicted and ATS system-neglected minor criminals back into workforce; predictive text, speech-to-text transcription, and voice and visual recognition for people with disabilities, gender neutral interview assessments, training facial recognition systems to work for a racially diverse pool and using attrition trends to identify top employee dissatisfaction areas are some of the examples of a combination of AI+ human intervention that is transforming workplaces. But most crucially, what all these workplaces have in common is a committed and sensitive leadership that pro-actively shapes these technologies to eliminate bias.

Leadership for the future

In a PwC survey, 76% organisations agreed that diversity and inclusion is a top priority yet only 26% organisations have D&I goals for leaders. Without goals against which performance can be measured, a diversity effort is bound to be ineffective or go unnoticed by stakeholders. Leaders can start by making a business case for diverse teams and build supportive policies and procedures accordingly.

An effective D&I programme must start by collecting data around the status of D&I efforts today. The next step should be building an inspirational strategy that is officially communicated to all stakeholders. Some of the key questions to answer at this stage are:

  • Assessing who could be employees/candidates adversely affected?
  • Do AI systems, if being used, further their exclusion? Are you solving the problem at the design stage?
  • Is there a team that will ensure that AI tools are being used for their intended function?
  • Are there channels for excluded groups to raise complaints? Are there individuals/teams to hear these grievances?
  • Is there a provision for regular fairness assessment of algorithms?
  • Are we investing in the security of AI systems?
  • Are all stakeholders a part of these communications?


Even with a solid diversity and inclusion programme in place, it could be months before an organisation sees the actual impact. But this shouldn’t be considered as a deterrent and should, in fact, be seen as an opportunity to create a long-lasting change. A sustainable diversity movement might take time to build but when solidified, it can work wonders for an organisation’s culture, innovation, efficiency, and growth.