AI in human resources: Technologies and HR applications

Artificial intelligence poses key challenges for human resources in companies: On the one hand, the uncontrolled use of the technology needs to be structured and skills realigned. On the other hand, AI itself should become an operational and strategic tool for optimized HR processes. This concerns, for example, the reading of CVs, employment contracts or applications as well as data-driven personnel planning and decision-making. However, in order for human resources to remain human, special care and consideration of ethical principles are required.

The most Important in a Nutshell

• Artificial intelligence is changing the use of information, the importance of various skills and requires companies to make structural adjustments. This means that the technology falls within the remit of HR.
• The central innovation driver for the predominantly text-based HR processes is Generative AI. However, technologies for predictive data analysis and intelligent document processing are also frequently used. Concrete application possibilities include:
• Automation of routine tasks - AI can automate time-consuming, repetitive tasks such as CV pre-selection or appointment scheduling.
• Sentiment analyses and employee satisfaction - AI analyzes employee feedback (e.g. surveys, internal communication) to gain insights into satisfaction and sentiment within the company.
• Improved decision making - - By providing a comprehensive information base, AI systems can promote more objective and unbiased decisions.
• Data-based talent acquisitionAI can be used to conduct targeted applicant searches. Intelligent search algorithms identify potential talent based on specific qualifications and soft skills.
• Personalized employee experienceAI systems can analyze individual employee preferences and provide tailored learning and development opportunities as well as benefits.
• The GDPR, works council co-determination rights and the European AI Act, which ensures compliance with ethical principles, apply.

Possible applications in the entire HR area

AI technologies, which are used to generate and process natural language, offer the potential to innovate all areas of modern HR. On the one hand, this is due to the predominant text data format and the heavily communication-based processes. On the other hand, mathematical key figures can also be used by management to make forecasts, plan resources and optimize capacity utilization.

Current Statistics show that 35 % of companies worldwide have already implemented AI technologies, while a further 42 % are actively evaluating and strategically developing their potential.

HR automation on a text basis

AI has long since found its way into everyday HR processes, often unnoticed and initially informally. Particularly in areas that are heavily influenced by text data, applications such as generative AI are quietly but steadily unleashing their transformative potential. In recruiting, for example, the automated analysis of CVs or the generation of suitable candidate assessments is increasingly taking the time pressure out of applicant selection. The rapid creation of professional rejection texts or the agile planning of job interviews using language models also relieve HR teams in a targeted and unobtrusive manner. The use of such technologies often starts out of pure curiosity: tools such as ChatGPT quickly find their way into everyday working life without any immediate comparison with compliance requirements or IT strategies.

However, this decentralized use also harbours risks. As soon as data, often including highly sensitive personal information, migrates to unsecured platforms, there is a risk of shadow AI: a parallel process that operates far from organizational control and can have data protection consequences. To counteract this development, the spontaneous progress initiated by individuals must be steered in an orderly fashion. This requires close coordination between HR and IT in order to develop robust security concepts, clear process guidelines and stringent compliance. Only if the use of AI is structured and consciously integrated into the HR strategy can the full value of the technology be exploited - from improved employee satisfaction analyses to the flexible creation of training and onboarding materials.

The influence of large language models is not limited to the digital space. Manuals and the documentation of processes give them comprehensive access to the process level. When used in a targeted manner, these insights can improve collaboration in a structured way.

Knowledge as a human resource

Artificial intelligence, in particular large language models (LLMs), makes it possible to organize and use knowledge in a way that goes beyond conventional approaches by dissolving communicative boundaries between departments and in natural communication with computer systems. The ability of these technologies to recognize, link and provide different forms of knowledge - whether explicitly in reports and documents or implicitly in experiences and processes - is particularly important. In HR processes, the practical effect can be seen in several places: The externalization of knowledge, for example through the automated conversion of verbal meeting minutes or employee observations into structured texts, reduces manual effort and makes valuable insights accessible to the organization. Combination takes place when AI integrates HR data sources such as feedback forms, onboarding documents and employee interviews to create reports that summarize strategically important insights. This data aggregation reduces errors and provides a sound basis for decisions on recruitment, personnel development measures or structural adjustments.

Internalization is particularly relevant: documented and generated content is provided by AI-supported tools such as chatbots at a low threshold in familiar working environments. One example is the provision of guidelines or training materials for new employees directly in their workflows. This low-threshold access ensures that knowledge is not only collected, but can be used immediately. The greater challenge, however, remains socialization, as direct interpersonal exchange can only be supported by AI to a limited extent. Here, HR managers need to play an active role in using AI as a means of preparing and structuring the basis for discussion. The integration of such processes not only changes how knowledge is organized in companies, but also reduces its dependence on individual availability. This makes personnel knowledge more stable, even when employees leave, and knowledge as a resource becomes more strategically controllable. HR departments that rely on AI-supported knowledge management at an early stage can actively shape this change and take on a role as central information managers.

Natural language meets structured data

In addition, LLMs now also enable the applicability of an approach that has long been considered too unwieldy or even impractical: knowledge graphs are used for the structured representation of objects and concepts within an organization. This digital network consists of nodes, edges and labels. Nodes can be employees, qualifications or departments, for example. Edges represent their relationships, such as the fact that an employee has a qualification and works in a certain department. Labels form typologies of edges and nodes, for example, whether they are persons or departments. A complete, spatially represented knowledge graph enables individual question and search systems that incorporate the knowledge of the entire organization. This not only seems complicated, it actually is - especially when it comes to the manual maintenance of the underlying graph database. In addition, many of the projects were nipped in the bud due to poor data quality, the isolation of knowledge silos and excessive costs.

But what makes the current application more difficult is the question of typical competencies of LLMs. These also capture information from unstructured formats, break down knowledge silos with comprehensive connections and can support the enrichment of a knowledge graph with their generative and logical capabilities. Instead of graph-based query languages, interaction is possible in natural language, which is then translated - thus facilitating access for non-technical users. But significant improvements are also possible the other way round: thanks to their basic structure of spatial relations, knowledge graphs form an ideal basis for training LLMs and thus providing them with basic knowledge. This is of crucial importance for the subsequent behavior of the models, which react sensitively to low data quality or distortions in the data structure. By systematically building the graph under human control and based on verified facts, confidence in the generated AI output ultimately increases. This gives HR managers complementary technologies a sound basis for our own company-specific decisions. And these in turn affect another human resource.

Systematic Skill mapping is becoming the core task of HR

As part of a progressive "Skills-first economy" competence management becomes a task that needs to be solved in ever shorter iteration steps. AI-based technologies, especially LLMs, play a transformative role here by enabling semantically sound analyses that allow an understanding of the real meaning structures behind mere keywords. The ability to recognize and contextualize implicit sub-skills in text data enables a detailed and reliable structuring of the available skill portfolio. So-called skill mapping based on AI is not a static process, but a continuously learning method: LLMs analyze extensive internal data sets - from CVs and project reports to internal communication data - and use them to create dynamic skills maps. These penetrate the deeper meaning behind superficial keywords by capturing interlinked skills such as negotiation skills, time management or crisis management in their overall context. This creates a semantically rich and resilient basis for skills management.

The strategic benefits of such skill mapping processes are particularly evident in the ability to work out the status quo of skills distribution in the company in detail. Automated SWOT analyses based on the data extracted by LLMs enable the systematic identification of existing strengths, weaknesses, opportunities and threats in the company's own skills structure. At the same time, these analyses enable the precise localization of skills gaps that can be addressed through targeted measures, for example through further training initiatives, internal retraining or new hires. The application of this "skills intelligence" thus allows reactive adaptation to current requirements, but at the same time lays the foundation for proactive skills development: the ability of LLMs to calculate various scenarios in advance based on certain criteria helps here. As soon as HR teams are aware of an internal need for skills, they can start systematically analyzing the job market.

Large language models make recruiting more precise

Significant advantages arise in particular when screening applications, matching candidates with specific job requirements and communicating with potential talent. Here, too, the ability of LLMs to recognize semantic nuances and independently prioritize decision bases helps. However, the introduction of these technologies also requires a critical examination of ethical issues. This concerns privacy, for example, Data Privacypossible discrimination and distortions in the algorithms. Effective implementation requires not only technical expertise, but also a comprehensive understanding of the interaction between humans and AI.

Practical application examples in recruiting:

Personalized communication - Automated, but personalized by AI, candidates can receive specific follow-ups or further information on the advertised position.

Screening of applications - LLMs analyze and structure applications according to relevant categories such as skills, education and professional experience. This speeds up the screening process considerably and optimizes the pre-selection process.

Automatic matching of profiles - Through NLP techniques allows job descriptions and applicant profiles to be compared and the best matches to be identified - more precisely than conventional keyword-based systems.

Initial contact with candidates - Chatbots answer questions, carry out qualification checks and thus create more space for the strategic focus of human recruiters.

Interview coordination - In some cases, LLMs can also facilitate scheduling by integrating calendar systems, taking candidate availability into account and generating suggestions for appointments.

Study proves the increase in performance through LLMs in recruiting

A Study by Japanese researchers investigates the use of Large Language Models (LLMs), such as GPT-4, to automate application processes in IT organizations, significantly outperforming manual screening methods in terms of efficiency. The framework developed includes data preparation, removal of personal information, assessment and summarization of CVs, and the final decision based on assessments and summaries. With an accuracy that comes close to the judgment of experienced HR experts, the system enables a prioritized selection of the best candidates. The flexibility of the method allows adaptations to industry-specific criteria, e.g. technical competencies or soft skills, and takes into account different industries such as IT, marketing or education. Despite high efficiency and potential scalability, the need to collect more diversified CV data for further development and generalizability of the approaches is pointed out.

Comprehensive transformation of the onboarding process

Once companies have analyzed their internal skills requirements and identified the right candidate, a structured, phased onboarding process begins. Each phase is transformed by AI, combining automation, data-based analysis and adaptive processes. It should be noted that the subsequent transition from onboarding to long-term workforce planning is now seamless, including dynamic retraining or upskilling in a changing skills context.

The key technologies that have crystallized are above all AI-based chatbots and Intelligent Document Processing (IDP) out.

1. preboarding: structuring formalities

Preboarding lays the foundation for a positive experience for new employees by completing all formal and organizational tasks before their first day at work.

Chatbots in preboarding:

• Answering frequently asked questions about document requirements, working hours, parking options or other initial concerns, taking the pressure off the HR department.
• Chatbots proactively inform new employees, for example by explaining which documents are missing or which deadlines need to be met.

Intelligent document processing in preboarding:

• IDP systems scan and validate uploaded documents such as tax forms, proof of identification or contracts and automatically recognize missing information.
• Automatic extraction of relevant data from submitted forms and feeding into internal systems such as payroll accounting or personnel files.

Practical example:
A new employee uploads their signed employment contract. The AI system checks the contract and transfers the relevant data to the HR system, while a chatbot communicates the processing status and answers questions about the contract.

2. orientation: providing information

During the orientation phase, new employees are given the basics to understand their role and find their way around the company. Chatbots and intelligent document processing ensure that information is quickly accessible and individually prepared.

Chatbots in orientation:

• Chatbots offer individual support by answering questions on organizational topics such as the use of internal systems, guidelines or office equipment.
• They guide you through personalized checklists, e.g. for IT access, initial training or organizational tasks such as registering with the company cell phone portal.

Intelligent document processing in orientation:

• Automatic analysis and compilation of important documents such as safety guidelines or procedural instructions, tailored to the employee's specific role.
• Intelligent document processing can extract and analyze data from feedback questionnaires, progress reports or induction plans and derive patterns from them. For example, aspects such as job satisfaction or challenges in team integration can be identified.

Practical example:
A new employee asks a chatbot how to use the time recording system. While the chatbot provides the instructions, the AI has already analyzed relevant manuals and provides the most important section directly in the chat.

3. integration: sustainable integration

The integration phase determines how well new employees are integrated into the corporate culture and their teams. AI-supported technologies can make a contribution to networking here, but this phase is characterized by human interaction.

Chatbots in the integration:

• Chatbots accompany the employee during the first few weeks, provide information on open tasks and offer support with common challenges.
• Connected to centralized knowledge repositories, chatbots promote the rapid closure of knowledge gaps across departmental boundaries.

Intelligent document processing in integration:

• Automatic evaluation of feedback questionnaires, performance measurements and feedback to identify progress and potential obstacles.
• Intelligent document processing ensures that personal documents relating to training courses and certificates are automatically recorded, checked and archived.

Practical example:
After the first few weeks, a chatbot asks the new employee for feedback on the induction process. The AI analyzes these answers together with progress documents and suggests measures to the team leader, e.g. specific training or integration into another project.

AI software for the HR sector

The AI software Konfuzio shows how the use of AI can promote efficiency, precision and strategic alignment in HR. As a central solution, Konfuzio integrates seamlessly into existing HR systems and addresses key pain points such as error-prone manual data processing, inefficient personnel planning and fluctuation risks. Using state-of-the-art AI technologies, the platform enables automated processes, data-driven decisions and a significant reduction in administrative effort. This makes Konfuzio an effective tool for relieving HR departments and strengthening their strategic importance within companies.

• Automated document processing
  Konfuzio facilitates the processing of HR documents such as employment contracts, CVs, vacation requests and payslips. With the help of HTR and OCR technologies, handwritten or printed content can be automatically captured, classified and archived - legally compliant and efficiently.
• Optimization in the onboarding process
  The automated answering of standardized inquiries via AI chatbot accelerates onboarding, improves the employee experience and shortens the time-to-productivity of new talent.
• Precise workforce planning
  Using data-based forecasts, Konfuzio supports demand and resource planning and helps to identify and avoid over- or understaffing at an early stage.
• Efficient recruiting
  AI-supported matching algorithms optimize the recruiting process by accelerating candidate screening and the planning phase This reduces the time-to-hire - without compromising the selection of the best talent.
• Strategic personnel controlling
  Automated data extraction and processing provide up-to-date key figures for more precise decision-making, more transparent HR controlling and well-founded fluctuation forecasts.
• Integration and flexibility
  Konfuzio is seamlessly integrated into HR systems via REST API. The software also meets the highest security requirements thanks to flexible hosting (cloud, on-premises or hybrid) and GDPR-compliant data processing.

Conclusion

Due to its high level of influence on information usage and human behavior, artificial intelligence exceeds the remit of IT and is becoming a structural issue for human resources. Those responsible are faced with the task of coordinating the realignment of the distribution of competencies in companies - between employees and departments as well as between humans and machines. The ability of AI systems to sift through large amounts of data and recognize patterns can support this optimization. At the same time, administrative tasks relating to recruiting, onboarding and contract data management can be relieved. For the integration of the technology to succeed, however, precise clarification with internal compliance, data protection and ethical principles is necessary. HR and IT can work together on an interdisciplinary basis to build up the necessary AI expertise, eliminate reservations and increase acceptance within the company - an important prerequisite for the successful use of AI.

Are you interested in artificial intelligence and its potential applications in HR? Please send us a message to get advice from experts.