Blog 02 - AI in Recruitment: Smarter Hiring or Algorithmic Bias?

AI in Recruitment: Smarter Hiring or Algorithmic Bias?

Recruitment has always involved difficult choices. Organisations must attract suitable candidates, evaluate large amounts of information and make decisions that can affect both organisational performance and an individual’s career. Artificial Intelligence (AI) is increasingly being introduced into this process to screen CVs, match candidates with vacancies, analyse applications, support assessments and communicate with applicants.

The attraction is obvious. AI can process information quickly and consistently, potentially reducing the administrative burden on recruiters. However, recruitment is not simply an efficiency exercise. It is also a process involving fairness, diversity, transparency, trust and human judgement.

The key question is not whether AI can make recruitment faster. It is whether AI can help organisations make recruitment better and fairer.

Why Organisations Are Turning to AI Recruitment

Recruitment can be expensive and time-consuming, especially when organisations receive hundreds or thousands of applications for a small number of vacancies. AI-supported systems can help recruiters identify keywords and skills, rank candidates against job criteria, communicate through chatbots and automate repetitive stages of the hiring process.

The OECD (2023) identifies several potential advantages of AI in labour-market matching. These include efficiency and cost savings, better matching through larger applicant pools, improved candidate experience and the possibility of reducing some forms of human bias. AI can also provide a more standardised experience by applying the same predefined criteria to many candidates rather than allowing every applicant to be judged through a completely different human process.

This connects with Strategic Human Resource Management. Recruitment is not only about filling vacancies; it is about acquiring the knowledge, skills and capabilities the organisation requires. Bratton and Gold (2017) emphasise that recruitment and selection are fundamental people-management activities because poor hiring decisions can affect performance, retention and organisational capability.

From this perspective, AI can be valuable when it helps HR professionals identify relevant talent more efficiently. But efficiency alone cannot be the final measure of a good recruitment system.

Candidates Are Not Fully Comfortable with AI Making Hiring Decisions

Figure 1. Senior decision-makers’ discomfort with allowing AI to perform people-management tasks. Source: CIPD (2023), Using AI responsibly in people management.

The figure above shows why recruitment deserves particular attention. In the CIPD study, 67.7% of senior decision-makers without HR responsibility reported some level of discomfort with allowing AI to shortlist candidates for interview. Their discomfort was even greater when AI was used for high-stakes decisions such as identifying or dismissing underperforming employees (CIPD, 2023).

This suggests that people distinguish between using AI as an administrative tool and allowing AI to exercise significant influence over a person’s career. Scheduling an interview or answering a routine question may feel relatively low-risk. Rejecting a job applicant is different because the consequences are much greater.

This issue can be understood through the concept of procedural justice. Fairness is not only about whether the final outcome is favourable. People also judge whether the process used to reach that outcome is transparent, relevant, consistent and respectful. Rigotti and Fosch-Villaronga (2024) highlight the importance of fairness in AI-enabled recruitment and show that fairness cannot be treated as a purely technical issue. It also involves legal, organisational and applicant perspectives.

Can AI Remove Human Bias?

One of the strongest arguments in favour of AI recruitment is that human recruitment is already imperfect. Recruiters and managers may be influenced by unconscious bias, first impressions, stereotypes, similarity attraction and inconsistent interviewing practices. The OECD (2023) notes that AI may introduce greater consistency and may help reduce some forms of human bias when systems are carefully designed.

However, AI does not appear from nowhere. Algorithms are created by people and trained using data generated by organisations and societies. If those data contain historical patterns of inequality, the system may learn and reproduce them.

Figure 2. Views on whether AI reduces or increases recruitment bias. Source: CIPD (2023), Using AI responsibly in people management.

The CIPD findings show how divided this debate remains. 38.6% believed AI could reduce bias, 31.2% believed AI had roughly the same level of bias as a person, and 30.2% believed AI could increase bias (CIPD, 2023). There is therefore no simple conclusion that “AI is objective” or that “humans are fairer”. The outcome depends on the data, design, criteria, governance and human oversight surrounding the system.

Bias in → Bias out?

If an organisation historically hired a narrow type of candidate and then trains an AI system to identify people who resemble previous successful hires, the technology may reproduce the organisation’s historical preferences rather than question them. The danger is that bias can become faster, more consistent and less visible.

A Real Example: Automated Screening and Age Discrimination

A useful real-world example comes from the United States. In 2023, the U.S. Equal Employment Opportunity Commission (EEOC) announced a US$365,000 settlement in a case involving iTutorGroup. According to the EEOC, the company’s online application software had been programmed to automatically reject female applicants aged 55 or older and male applicants aged 60 or older. The EEOC stated that more than 200 qualified applicants were rejected because of age (EEOC, 2023).

This example demonstrates an important principle: organisations remain responsible for employment decisions even when those decisions are supported or automated by technology. Delegating part of a recruitment process to software does not remove the organisation’s responsibility to ensure the process is fair and job-related.

The EEOC (2024) also explains that AI may be used in recruitment for activities such as screening CVs for keywords or experience, evaluating recorded video interviews and encouraging candidates to apply through chatbots. This illustrates how deeply automated systems can now enter the employment relationship before an applicant even speaks to a human recruiter.

Fairness Is Also About Candidate Experience

A technically accurate recruitment system can still damage an organisation if candidates perceive it as unfair or impersonal. Recruitment is one of the first experiences a potential employee has with an organisation. Therefore, the recruitment process contributes to the employer brand and shapes candidates’ expectations of how people are treated inside the organisation.

Horodyski (2023) found that applicants can perceive AI positively and may view it as useful and easy to use. However, other research indicates that reactions depend strongly on how the technology is used, particularly the degree of human involvement and the stage of the selection process. Gonzalez et al. (2022) found that people responded more favourably when AI augmented human decision-making rather than replacing it completely.

This is strategically important. A company may save time by automating rejection decisions but lose talented candidates if applicants believe the organisation is unwilling to provide transparency or human consideration.

A recruitment system should not only identify good candidates. It should also create a process that good candidates are willing to trust.

The Human + AI Approach

The strongest approach may therefore be neither fully human recruitment nor fully automated recruitment. A Human + AI model allows technology to assist with high-volume tasks while keeping human professionals responsible for interpretation, accountability and high-stakes decisions.

The OECD (2023) discusses the value of a “human-in-the-loop” approach, where human judgement remains between the automated system and the final outcome. Similarly, CIPD guidance recommends keeping people in charge, testing systems rigorously, auditing for bias and avoiding AI where the selection decision requires nuanced judgement that cannot reasonably be reduced to a formula.

A responsible AI-supported recruitment process should therefore include:

• Clear job-related criteria before the AI system is used.
• Diverse and representative data rather than blindly learning from historical hiring patterns.
• Human review of shortlists and important rejection decisions.
• Regular bias audits to identify unequal outcomes across candidate groups.
• Transparency so candidates understand when and how AI is being used.
• Accessibility and reasonable alternatives where automated tools disadvantage certain applicants.
• Accountability so responsibility remains with the organisation rather than being shifted to a technology vendor.

These practices also support inclusive recruitment. CIPD (2024) recommends structured and transparent selection processes that provide a fairer basis for comparison between candidates. Technology should reinforce these principles rather than undermine them.

A Global HRM Perspective

The issue becomes more complex for multinational organisations. Brewster et al. (2017) emphasise that international HRM operates across different institutional, cultural and legal environments. A recruitment technology designed in one country may be used to evaluate candidates from many different cultural backgrounds.

This creates several questions. Will the system recognise qualifications from different education systems? Does it interpret language differences fairly? Could it penalise candidates whose CV formats do not match the format used in the training data? Are attitudes towards automated decision-making the same across countries?

Global HR professionals therefore need to avoid assuming that one automated recruitment model can simply be transferred everywhere without adjustment. Responsible global HRM requires technology to be tested within the cultural, legal and organisational context in which it is used.

🎥 Recommended Video

World Economic Forum – Can You Still Get Hired in the Age of AI?

This video explores how AI is changing hiring and why maintaining the human element remains important in recruitment.

▶ Watch on YouTube

My Reflection

Before examining this issue, I viewed AI recruitment mainly as an efficiency tool. If a system can review hundreds of applications in seconds, it seems logical that organisations would want to use it.

However, this topic has shown me that recruitment quality cannot be measured only by speed. A recruitment decision can be efficient but still be unfair. It can be data-driven but still be based on poor data. It can be consistent but consistently reproduce the wrong pattern.

I also realised that human recruitment should not automatically be considered the “fair” alternative. Humans have biases too. The real challenge is therefore to design a process where the strengths of technology are combined with the strengths of human judgement.

In my view, AI should be allowed to support recruitment but should not become an unquestioned authority over recruitment. HR professionals must understand why candidates are being selected or rejected, regularly challenge the system’s outputs and remain accountable for the final decision.

AI can make recruitment faster. Human responsibility is what can make it fair.

💬 Join the Discussion

Would you be comfortable if an AI system decided whether you received a job interview?

Do you think AI can make recruitment fairer than human decision-making, or could it simply hide existing bias inside an algorithm? Share your view in the comments.

References

Bratton, J. and Gold, J. (2017) Human Resource Management: Theory and Practice. Basingstoke: Palgrave Macmillan.

Brewster, C., Sparrow, P., Vernon, G. and Houldsworth, E. (2017) International Human Resource Management. 4th edn. London: CIPD.

CIPD (2023) Perceptions of fairness: using AI in selection and recruitment. Available at: https://www.cipd.org/uk/knowledge/bitesize-research/fairness-ai-recruitment/ (Accessed: 10 August 2026).

CIPD (2023) Using AI responsibly in people management. Available at: https://www.cipd.org/en/views-and-insights/thought-leadership/insight/ai-people-management/ (Accessed: 10 August 2026).

CIPD (2024) Inclusive recruitment: Guide for people professionals. Available at: https://www.cipd.org/en/knowledge/guides/inclusive-employers/ (Accessed: 10 August 2026).

EEOC (2023) iTutorGroup to Pay $365,000 to Settle EEOC Discriminatory Hiring Suit. U.S. Equal Employment Opportunity Commission. Available at: https://www.eeoc.gov/newsroom/itutorgroup-pay-365000-settle-eeoc-discriminatory-hiring-suit (Accessed: 10 August 2026).

EEOC (2024) Employment Discrimination and AI for Workers. U.S. Equal Employment Opportunity Commission. Available at: EEOC resource (Accessed: 10 August 2026).

Gonzalez, M.F. et al. (2022) ‘Allying with AI? Reactions toward human-based, AI/ML-based, and augmented hiring processes’, Computers in Human Behavior, 130.

Horodyski, P. (2023) ‘Applicants’ perception of artificial intelligence in the recruitment process’, Computers in Human Behavior Reports, 11, 100303. doi:10.1016/j.chbr.2023.100303.

OECD (2023) Artificial Intelligence and Labour Market Matching. OECD Social, Employment and Migration Working Papers. Paris: OECD Publishing. Available at: OECD report (Accessed: 10 August 2026).

Rigotti, C. and Fosch-Villaronga, E. (2024) ‘Fairness, AI & recruitment’, Computer Law & Security Review, 53, 105966. doi:10.1016/j.clsr.2024.105966.