The study examines whether AI can help a structured and transparent evaluation process while retaining human oversight over final recruitment decisions. These documents are usually examined manually, which makes the process time-consuming, subjective, inconsistent, and subject to cognitive biases like confirmation bias, affinity bias, and the halo effect.

The study adopts a Design Science Research (DSR) approach consisting of the Awareness, Suggestion, Development, and Feedback phases to design the Hybrid Human–AI Qualitative Evaluation Framework (HHAQ-EF). The Awareness phase included eight semi-structured expert interviews, which identified challenges including reliance on individual judgement, inconsistent evaluation criteria, time-intensive manual review, limited use of AI for qualitative assessment, and the absence of a structured evaluation method for qualitative recruitment texts. These findings influenced the design objectives and the selection of a hybrid Human–AI approach.The final framework is classified into four layers: input and preparation, AI-supported text analysis, human review and decision-making, and transparency and governance. AI-generated insights, such as semantic relevance, sentiment, tone, readability, and bias indicators, are combined with human supervision to provide more consistent and standardised evaluations. The approach was demonstrated with a simulated evaluation of eight candidates and reviewed by practitioners, including open-ended survey responses. Participants saw the framework as a decision-support tool, emphasising the importance of human monitoring.

The thesis statement is partially confirmed. The findings indicate that AI can support HR professionals in evaluating qualitative recruitment texts by improving efficiency, consistency, and transparency through a structured evaluation process while supporting human decision-making and increasing awareness of potential cognitive bias. However, AI should complement rather than replace human judgement, and further empirical validation in real recruitment settings is required to assess its effect on bias reduction.