AI tools in selection: How do we know we're on the right track?

The integration of artificial intelligence in recruitment is no longer a matter of the distant future but of strategic thinking in the present. For “We Asked the Experts”, Todor Ranchev, Co-Founder at TalentSight, shares his perspective on the role of AI in recruitment processes — when it adds real value, where the human factor remains irreplaceable, and what steps organiсations can take to properly implement these processes in line with their business needs.

What have we learned so far from the failures and successes of AI tools in recruitment — what mistakes should we avoid to prevent them from becoming a 'black box' for talent?

The integration of artificial intelligence in recruitment is no longer a matter of the distant future but of strategic thinking in the present. For the column “We Asked the Experts”, Todor Ranchev, Co-Founder at TalentSight, shares his perspective on the role of AI in recruitment processes — when it adds real value, where the human factor remains irreplaceable, and what steps organisations can take to properly implement these processes in line with their business needs.

What have we learned so far from the failures and successes of AI tools in recruitment — what mistakes should we avoid to prevent them from becoming a 'black box' for talent?

When we talk about AI in recruitment, there are several key processes where technology can be effectively implemented. First and foremost, it's important to emphasise that AI tools yield the best results where there is a large volume of data. Recruitment is precisely such a field — with a huge and ever-growing volume of constantly updated data.

At first glance, recruitment seems like the perfect environment for applying AI. However, there is one important distinction. Unlike areas like finance, where decisions are based almost entirely on numbers and objective values, recruitment is centered around people. That’s why I believe AI should not be the final decision-maker, but rather serve as an assistant in the process. This doesn’t mean AI has no place in recruitment — rather, it should be implemented where it supports professionals without replacing human judgment.

Here are a few examples of recruitment processes where AI can be useful, along with specific AI tools:

• Market research (LinkedIn Talent Insights)
• Sourcing (LinkedIn Recruiter, TalentSight)
• Pre-screening (Carv, HireFlix)
• Interviews (Asendia, Carv)

How do we know when we need to implement an AI tool?

There are two main benefits to implementing AI tools in recruitment. In the context of recruitment agencies, the main advantage is the potential to increase revenue, while for internal teams, the focus is more on cost optimisation.

Regardless of the type of organisation, if you have an open position for a Talent Acquisition Specialist, that’s a clear signal the company is looking for ways to strengthen its recruitment process. In such a situation, it’s worth asking:

• What is currently preventing us from achieving our desired results?
• In which specific processes would we like to improve efficiency?

I can say with high confidence that there are already AI tools that meet the needs of your exact process. But they won’t find you on their own — you need to seek them out and experiment.

Then comes the real evaluation phase: Can we, by implementing the right AI tool and training our existing team, achieve the same (or better) results — without hiring an additional person? I believe we can, and there are real examples of this happening.

What should we consider when choosing an AI tool for recruitment — from the quality of the data the algorithm is based on, to the transparency of its processes?

When working with an AI tool, it’s important to understand the history and origin of the datasets the model was trained on. Equally important is comparing this information to the current context and market in which we operate.

A simple example — if an AI tool was developed in East Asia and trained with data reflecting local specifics and user behavior there, it’s possible its results won’t be as relevant or effective in Bulgaria as they would be in Thailand, for instance.

Context matters — and that’s key when selecting and applying any AI solution.

How can we measure the success of AI tools in the recruitment process? What specific indicators and KPIs confirm that investment in automated solutions leads to better hiring quality, reduced bias, and accelerated processes?

The KPIs we use to determine whether an AI tool is worth the investment must align with core business metrics. For example, one of the key metrics in recruitment is time-to-hire. If the hiring time decreases after implementing a specific tool, that’s a clear indicator of positive impact. Of course, this metric can be influenced by many external factors. That’s why the best and quickest way to assess the effectiveness of an AI tool is through smaller, clearly measurable indicators. Examples include:

• How many candidates each recruiter brings into the process weekly.
• How many messages are typically needed to initiate a conversation with a potential candidate.
• What is the response rate after the initial outreach.

These metrics can vary depending on the specific recruitment process, but it’s important to track them both before and after the AI tool is implemented to make a real comparison and assess impact.

What does the perfect balance between AI automation and human expertise in recruitment look like?

Each recruitment process has its nuances and sub-steps, but to simplify things, let’s focus specifically on the sourcing process. Personally, I believe AI’s role should be to enhance and scale the sourcing process without sacrificing the effectiveness and control exercised by the recruiter. In my view, the best balance between AI and humans in recruitment is achieved when AI is used to optimise administrative and repetitive tasks — such as initial sourcing, starting conversations, profile analysis, pre-screening, note-taking, and interview analysis, among many others. This frees up the recruiter’s time to focus on the most valuable part of the process — building relationships with candidates. And most importantly — the final decision remains in human hands.