Many students in Germany face significant challenges in choosing a career path after graduation. Only a fraction choose to pursue careers in natural sciences or IT, as many lack personal access or are inadequately supported through their school, employment agencies, or parents. Students from socio-economically disadvantaged families are particularly affected, especially those with a migration background, who often have limited opportunities due to their different linguistic backgrounds, exposure to other educational systems, and monetary aspects.

Since 2014, the Hacker School gGmbH has been pursuing the goal of inspiring socio-economically disadvantaged students to learn programming as a hands-on skill, thereby simultaneously combating educational inequality and the lack of young talents in the IT industry sector. The federally funded research project TALAINTED, carried out in cooperation with the Institute for Applied Artificial Intelligence at Stuttgart Media University, aims to translate this concept into a multilingual online format using virtual AI avatars. This format allows for unlimited repetition and learning in small steps thanks to its independence of time and place.
This study, conducted over 5 days in late May/early June 2024 in Hamburg, Germany (schools with a low social index) and Tübingen, Germany (schools with a high social index), consisted of semi-structured qualitative interviews with 5 teachers and 79 students in grades 6-8, who were between 12-14 years old. Each interview lasted between 10 and 15 minutes.

Preliminary results show that the AI-supported online course, through its game-based structure, increases students' motivation and interest in IT careers, especially in coding as a professional skill, regardless of their prior knowledge in the field. The creative teaching approach of different programming languages at different levels of difficulty reflects the reality of students' lives, as many of them work with computers in their spare time. In addition, the course promoted a continuous improvement in participation, attention, and focus on the lesson compared to normal face-to-face formats, especially due to the adaptability of the learning content in terms of speed, unlimited repetition, and multilingual on-demand availability.

The study shows that knowledge creation and retention is deeper and more focused because of the small-step nature of the course elements, the ease of use through visual demonstrations, and the overarching storytelling. Cognitively low-threshold videos help overcome difficulties with vocabulary and attention span. In summary, the TALAINTED course promotes adaptive learning strategies and encourages autonomous knowledge transfer at different levels.

Keywords: AI, AI education, AI avatars, game based learning, video based learning.