Almost everyone is talking about artificial intelligence today. Boards, founders, CEOs and department heads. Technology companies, consultants, industry media, and even people who only a year ago were not quite sure what AI actually meant.
It is not hard to see why. AI promises something organisations have been looking for for a long time: more efficient work, fewer repetitive tasks and greater process scalability. In recruitment, that promise is especially tempting because it touches one of HR’s most practical problems: time and volume.
From the employer’s perspective, the main reason candidates do not receive feedback is lack of time and too many applicants. This was indicated by 59% of companies in eRecruiter’s 2025 report.¹ According to another report, the average time between an application being submitted and opened by a recruiter is 9 days, while 28% of applications are never opened at all.²
So what does AI offer? More efficient pre-screening. Faster contact with candidates. Automatic collection of basic information. And more time for recruiters to focus on work that requires context and experience.
At the level of the promise, everything sounds simple. But beyond the promise, there is implementation. And there is organisational reality. That is where questions and uncertainty appear.
Is this legally compliant? How do we navigate GDPR, the AI Act, the legal department and internal policies? How do we prepare the team to work with a new tool? Will candidates accept it? Will recruiters be able to trust system recommendations? Will the tool actually take work off the team’s plate?
These are good questions. Especially because AI in recruitment is not yet as obvious a tool as an ATS, a calendar or company email. For many organisations, it is still a new element of the process, something that needs to be understood, familiarised and properly embedded in everyday work.
So let us look at the conditions that need to be in place for such an implementation to actually make sense. Without technological hype. But also without pretending that modern recruitment can be run while ignoring tools that can genuinely relieve people of part of their workload.
Start with the problem, not the solution
AI should not be implemented just because it is fashionable. That would be unwise. It may sound obvious, but in organisational reality it is easy to forget. Especially today, when many companies feel pressure to “do something with AI”.
But “something” is not enough. An implementation only makes sense when the company can name the problem it wants to solve.
Sometimes the problem is first contact that happens too late. Sometimes it is repetitive questions that take recruiters hours to verify. Sometimes it is a high-volume process with simply too many applications.
These are concrete problems. Only when we ask, “which part of our process is the bottleneck today?”, we can check whether AI is the right answer.
If an organisation wants to implement technology only to show that it is innovative, it makes little sense. But if the company knows where it loses time, what it struggles to deliver, or what puts the greatest strain on the team, technology can help. European analyses of AI in Business Support Processes follow a similar line of thinking. HR is identified as one of the areas where AI can have practical application even at an early stage of adopting this technology.³
Changes to the process
Once a company knows which parts of the recruitment process could be supported by AI, another very natural question appears. Will this mean too much change?
Especially when we are talking about a new tool and a new provider. On the organisation’s side, more steps appear: legal, IT, budget, GDPR, documents, team onboarding and internal approvals. All of this takes time. And time is often already in short supply in HR.
It is easy to think that such a project may simply be too demanding. That the organisational cost of implementing such a solution may be too high.
Let us be honest: implementation will not happen by itself. Someone has to define the scope, go through formal requirements, prepare the team and adjust the process. But does the entire burden really have to sit only with HR?
A good provider has gone through similar implementations before and understands the questions that usually arise from the departments involved in the process. This means HR is not working alone and does not start from scratch.
That does not mean there is no work on HR’s side. The right people need to be involved, information needs to be shared and internal procedures need to be followed. But HR should not have to carry out the entire implementation on its own. Its role is often more about ensuring a smooth flow of information between the organisation and the provider.
Well-implemented technology does not have to turn the entire process upside down. The point is for the technology to enter the existing process and support the parts the company has already identified as bottlenecks.⁴ The process remains similar. What changes is that some tasks can be done faster, more smoothly and with less strain on the team.
And here a paradox appears. In conversations about AI, one argument comes up often: “we do not have time for this kind of implementation right now”. But very often, that lack of time comes precisely from the fact that the team is overloaded with tasks that could be partly structured, automated or delegated to technology. As a result, the very problem AI could help solve becomes the reason not to implement it.
Does the law say “do not use AI”?
Another common set of questions for organisations implementing AI in recruitment concerns legal issues. What requirements do we need to meet? Is this legal? And will we be able to implement it?
HR teams regularly work with candidate data, which is subject to GDPR. On top of that, there are regulations concerning artificial intelligence. In the European Union, AI used in recruitment and candidate selection comes with specific regulatory obligations. This means organisations need to look carefully at issues such as data quality, documentation, transparency and human oversight.⁵
The law is not there to block the adoption of new technologies. It is more of a framework within which organisations need to operate so that technology can work safely, responsibly and predictably.
That is why several issues should be clarified early on. What data is being processed? For what purpose? On what legal basis? Who has access to it? What role does the system play? Where does the human enter the process? And who is responsible for the decision?
In practice, this means a conversation between HR, the legal department and the solution provider. The organisation needs to go through the data processing agreement, information clauses, process description, scope of responsibility and oversight rules.
These are standard elements of a well-prepared HR tool implementation. The difference is that with AI it is especially important to describe clearly what the technology does, what the human reviews, and where the boundary lies between system support and organisational decision-making.
Can the system be trusted?
The next questions appear quickly. “How do I know the system assessed the candidate correctly?” Will it miss a strong candidate? Will it recommend someone on weak grounds? Will it work like a black box?
The problem does not start when a company asks these questions. The problem starts when the tool cannot answer them. In recruitment, a final score alone is not enough. A good implementation should make it possible to check where the recommendation came from. Which criteria were taken into account. How the candidate answered the questions. Which conditions they met and which they did not. Where doubts appeared. And what needs to be clarified in a conversation with a human.
AI should not replace thinking. It should prepare better material for decision-making. That distinction changes the way we look at the technology. A company should not take the system “at its word”. It should be able to see the logic behind the recommendation, verify the result and decide whether it makes sense. Trust in AI comes from transparency, control and the ability to verify.⁵
Could AI take the human element out of recruitment?
In HR, this question is natural. Recruitment is about people. Their career decisions, ambitions, motivations, plans and doubts. It is difficult to talk about AI in recruitment without asking whether we might lose the human element along the way.
But it is worth asking the question more precisely: where in the process is the human truly needed most? Is it when they ask, once again, about availability, salary expectations, readiness for shift work or start date? Or is it when they need to understand the context, talk about motivation, clarify doubts, assess fit and take responsibility for the decision?
AI should not replace the human in recruitment. It should relieve them of tasks that are repetitive and time-consuming. The fact that a recruiter can do something manually does not mean they always should. The point is to give them more time for the moments where their presence truly improves the quality of the process.
AI is not neutral. What about humans?
A conversation about AI in recruitment cannot avoid the topic of bias. A poorly designed system can reproduce errors. It can reinforce patterns. It can appear objective while in practice deepening inequalities. This must not be downplayed.
At the same time, a recruitment process run entirely by humans has never been free from bias either. Humans bring experience, intuition and context. But they also bring fatigue, preferences, mental shortcuts and inconsistent standards of assessment. One recruiter may notice something another recruiter overlooks. One hiring manager may value a non-standard career path, while another may see it as a risk.
AI is not an automatic cure for this challenge. Adding technology to the process is not enough to make recruitment fairer. But it can help control the assessment process better, provided the process is based on clear criteria, comparable questions, documentation, the ability to verify results and human oversight.⁵ Then it becomes easier to check why a candidate was assessed in a particular way.
A well-designed AI-supported process may be easier to review than many decisions made purely on intuition. And in recruitment, the ability to review the process matters. Because when a decision concerns people, it needs to be justified.
The case for AI starts with everyday work
After looking at questions about process, law, the role of the recruiter and bias, it is worth returning to the basics. The strongest arguments appear when we look at the everyday work of a recruiter.
They start with candidates who need to be called. Messages that need replies. Basic information that needs to be collected. Processes where there are many applications and not enough time. That is where we can see whether AI can create real value.
AI can shorten the time to first contact, help verify basic requirements faster and organise information from conversations. It can also reduce the number of repetitive tasks that are necessary, but do not always require the recruiter’s full attention.
From the organisation’s perspective, this is not just convenience. It means greater team capacity, fewer delays and better use of people’s work. Today, recruiters are often stretched between candidate communication, administration, business alignment and reporting.
A recruiter does not create the most value when repeating the same questions again and again. They create value when they understand the context of the role, can speak with the candidate, identify motivation, support the hiring manager and help the organisation make a good decision. If AI takes over part of the repetitive work, it does not have to weaken recruitment. It can give recruiters more space for the part of the work that technology should not take over.
You do not have to start with a major change
AI implementation does not have to start with a large project. It is often more sensible to begin with one part of the process that genuinely hurts the organisation. For example, first contact with candidates, initial qualification, collection of basic information, or support in high-volume processes where recruiters most often repeat the same actions.
Such a pilot has one major advantage. It moves the conversation about AI from declarations to practice.³ There is no need to decide immediately whether technology will transform the entire recruitment function. It is enough to check whether it helps solve a concrete problem.
Does the process work better after the change? Do recruiters save time? Do candidates get a response faster? Are recommendations understandable? Do hiring managers receive better information? Does the team trust what it sees in the system? Is the number of repetitive tasks going down instead of up?
These are not exciting questions. But they are exactly the questions that determine whether the implementation makes sense.
Team onboarding is also important here. Even a good tool will not work well if people do not know when to use it, how to read its outputs and how to combine system recommendations with their own judgement.
That is why AI implementation should include not only technology configuration, but also user preparation. Not so that every recruiter becomes an expert in language models. But so that they know how to use the tool in everyday work and where the role of the system ends and human responsibility begins.
Do not “implement AI”. Check whether you have a problem AI can solve
Let us return to the beginning. Almost everyone is talking about AI today. And they will probably talk about it even more. But in recruitment, the point is not to implement technology just because the market is looking in that direction.
The point is to check whether AI can solve a concrete problem in the process. If the answer is “yes”, the conversation about AI stops being a conversation about fashion. It becomes a conversation about better recruitment.
Of course, implementation will require questions. About the process, law, data, responsibility, trust, the role of the human and team preparation. And that is a good thing. These are exactly the questions that separate a sensible implementation from a blind rush toward technology.
AI should not be a shortcut. It should be a way to remove from people the part of work that unnecessarily takes their time, energy and attention. Then the human does not disappear from the process. They appear where they are truly needed.
Footnotes:
¹ eRecruiter, Candidate Experience w Polsce 2025, eRecruiter, 2025.
² eRecruiter, Rekrutacyjne KPI 2025, eRecruiter, 2025.
³ StepUp StartUps Consortium, The Opportunity for AI to Enhance Business Support Processes, report published as part of the EU-funded StepUp StartUps initiative, 2025.
⁴ Department for Science, Innovation and Technology, Responsible AI in Recruitment, GOV.UK, 2024.
⁵ European Parliament and Council of the European Union, Regulation (EU) 2024/1689 of 13 June 2024 laying down harmonised rules on artificial intelligence, OJ L 2024/1689, 12.07.2024, Article 6(2), Annex III, point 4 and Articles 10–14.
