Firstly, let us agree on four fundamental attributes of AI in recruitment and in general use:
- – It wants to please, and AI does not challenge
- – It has no empathy because it deals only in fact or fiction
- – It has no intuition and cannot sense when something is missing
- – It is unbelievably powerful, but only as an enabler
The recruitment process contains eight major steps, and each one raises the question of where AI in recruitment can be a trusted partner and where the process needs stronger human input.
Everything below reflects what we have learnt first-hand while integrating AI in recruitment across multiple stages. Every time we introduce AI into a new task or workflow, we ask one important question that you should ask as well:
Will AI improve the process or improve its efficiency, or will it simply produce a weaker outcome just to save costs?
AI in recruitment should never be used if the outcome becomes worse than a human-delivered process, no matter how tempting the cost savings may appear.
1. Developing Role Benchmarks
Every role is unique to each organisation. The behaviours, competencies and cultural fit required are not generic. Proper role benchmarks must be created from real top performers or through management insights. Since AI in recruitment relies on generalised global information, it cannot understand your culture, your management style or the unique traits that make your workplace operate effectively.
Verdict: AI is not appropriate
2. Developing Position Descriptions
AI in recruitment shines when building position descriptions. Its ability to scan role benchmarks and management input allows it to create structured, concise, one-page position descriptions that include measurable KPIs. The process is fast, and the quality is often better than what most people can produce manually.
Verdict: AI is a must
3. Developing Job Advertisements
Most job advertisements receive the least amount of strategic attention. They often focus on the employer rather than the applicant. The result is an ad that fails to answer the critical question that quality candidates ask, which is “What is in it for me?”. AI in recruitment can transform this step by using global data, role benchmarks and position descriptions to create compelling advertisements that stand out and attract stronger applicants.
Verdict: AI provides a real advantage
4. Screening Job Applicant Resumes
This is one of the most popular applications of AI in recruitment. It saves time, but the question is whether it improves quality. AI can miss nuances, interpret things literally and overlook high-potential candidates who do not fit expected patterns. Many fantastic employees may be screened out before a human ever sees their resume. Time is saved, but the organisation may lose great talent.
Verdict: AI is a time saver, but not an organisational benefit
5. Conducting First Interviews
Some platforms now allow AI to conduct realistic first interviews. However, the consistency of these outcomes is still unclear. Ironically, the strongest argument for AI in this step is the generally poor interviewing skills seen across many hiring managers and sometimes even HR teams. This area may evolve quickly, so ongoing testing and small pilots make sense.
Verdict: The jury is out
6. Shortlisted Candidate Assessments
There is recent discussion about replacing psychometric assessments with AI in recruitment. Although the idea is appealing, reliable psychometric testing is built on scientific validation, norm groups and consistent algorithms. AI cannot yet produce consistent predictions of human behaviour and may present an inaccurate picture, especially when influenced by social media data that reflects how a person wants to be perceived rather than who they really are.
Verdict: AI is misleading and potentially dangerous
7. Conducting Second Interviews
The second interview is one of the most important steps in preventing costly hiring mistakes. At this stage, human judgment, emotional intelligence and contextual understanding are vital. Anyone using AI for final interviews is prioritising cost reduction over building high-performing teams.
Verdict: Never use AI for this step
8. Reference Checking
AI is currently used mostly to facilitate communication with referees rather than conduct conversations directly. These tools speed up the process and increase response rates, which makes them genuinely useful.
Verdict: AI provides benefits
Final Thoughts on AI in Recruitment
Throughout the attraction and selection process, there are stages where AI in recruitment is already highly beneficial, stages where it may become beneficial in the future and stages where it will never replace human insight.
The essential approach is to actively pilot new AI tools and measure their impact on organisational performance, employee engagement and overall hiring quality. Recruitment is and will remain a deeply human activity. Use AI in recruitment where it enhances outcomes, not where it undermines them. Do not outsource decisions to AI simply for cost savings, because the quality of your people is always worth more.
Written by Mark Purbrick
