Higher education plays a critical role in shaping human capital by equipping individuals with the essential skills needed to thrive in the modern and competitive world. However, higher education has been significantly disrupted by generative artificial intelligence, which challenges traditional education models. This disruption began with the introduction of ChatGPT in November 2022, which rapidly gained popularity for its natural language processing capabilities, followed by the development of various other generative AI chatbots (e.g., Microsoft Copilot, Google Gemini) and cost-effective reasoning models (e.g., Deepseek), making generative AI even more accessible to higher education students. At the same time, the labour market has begun to feel the impact of generative AI, making some jobs obsolete through displacement, reshaping existing roles through transformation, and improving efficiency through enhancement while simultaneously creating new jobs in emerging industries. While existing research provides a preliminary understanding of the potential and challenges of generative artificial intelligence for skills development and the labour market, it lacks strong empirical evidence.

Therefore, this paper aims to fill the gap in existing research by providing evidence on the evolving perceptions of students regarding generative artificial intelligence in higher education, with a focus on ChatGPT implications for skills development and the labour market. The empirical analysis is based on data collected through the Global ChatGPT Student Survey, initiated by the Faculty of Public Administration at the University of Ljubljana, Slovenia, which has so far been conducted in two iterations. The first round captured student perceptions of ChatGPT approximately one year after its introduction (between October 2023 and February 2024), while the second round was conducted two years after its introduction (between October 2024 and February 2025). A comparison between the two global student samples, with 23,218 participants in the first round and 22,963 in the second round, allows for an examination of how these perceptions have evolved over time.

The results show that initial perceptions of ChatGPT’s role in skills development were moderate, with limited recognition of its impact on professional writing. While some valued its brainstorming and proofreading abilities, doubts remained about its effectiveness in specialized skills. Views on artificial intelligence’s impact on the labour market were also mixed, with skepticism about job opportunities and employer expectations. By the second round, confidence in ChatGPT’s role grew, particularly in writing, idea generation, and technical skills, with users becoming more adept at leveraging artificial intelligence as a learning aid. More respondents acknowledged artificial intelligence-driven skills as a competitive advantage, though concerns about job displacement persisted. The comparative analysis highlights growing recognition of ChatGPT’s role in skill enhancement, especially in brainstorming and digital literacy, while professional writing remained a challenge, underscoring the need for structured artificial intelligence-assisted training. Discussions on artificial intelligence’s labour market impact evolved from skepticism to a more nuanced understanding, emphasizing adaptation and upskilling as essential for future employability.

Keywords: ChatGPT, higher education, skills development, labour market, student perception.