Abstract:

Growing use of Artificial Intelligence (AI) technologies and applications in almost all sectors of the economy and human activities creates demand for specialists on AI. Higher Education responded to this demand by establishing new AI focused undergraduate and graduate programs that attract the growing number of students. However, there is no formally defined AI Body of Knowledge and competences framework, similar to those defined for Computer Science, Software Engineering or Data Science. Many AI programs are created based on existing Computer Science, Data Science or Machine Learning programs and may not cover important knowledge areas.

This paper presents ongoing research and current results of applying combined bottom up (job market analysis) and top down (Analysis of top AI curricula) approach to defining job market demand for AI competences and knowledge and assessing what AI related disciplines comply with the market demand. The paper uses the EDISON Data Science Framework (EDSF) methodology [1] to analyse the job market data by matching the selected AI related vacancies against the multi-vector controlled vocabulary for AI related competences and knowledge topics that are identified based on the analysis of the leading AI curricula. The presented research investigates different text matching algorithms and produced reference dataset of the collected AI related job vacancies from indeed.com to help future research on defining community agreed AI competences and Body of Knowledge.

The initial set of AI-related competences and knowledge areas was identified based on the analysis of existing AI programs, in particular, the AI Bachelor and AI Master programs offered by the University of Amsterdam in Netherlands [2], which is recognized internationally. It was later verified and confirmed with the job market analysis what was the focus of this research.

The following is the enumerated list of competence groups and knowledge areas identified from existing curricula analysis (not specific relevance order at this stage):
CKAI_01 Knowledge representation and reasoning.
CKAI_02 Automated planning and scheduling.
CKAI_03 Machine learning.
CKAI_04 Natural language processing.
CKAI_05 Machine perception.
CKAI_06 Computer vision.
CKAI_07 Speech recognition.
CKAI_08 Robotics.
CKAI_09 Affective computing.
CKAI_10 Deep learning.
CKAI_11 Information retrieval.
CKAI_12 Computer science.
CKAI_13 Causality.
CKAI_14 Data mining.
CKAI_15 Commonsense knowledge.
CKAI_16 Intelligent agent.

The main challenge was to create a reference dataset that could serve as a controlled vocabulary for text analysis algorithms, in conditions that there is no formal definition of the AI competence framework and Body of Knowledge. It was correlated with the curricula and academic subjects description, and also active use of the ChatGPT4.0 for identifying knowledge topics inter-relations. In the process of research, the reference dataset has been revised multiple times to ensure the best possible competences identification in analysing vacancies certainty and minimising data bias.

References:
[1] The Data Science Framework, A View from the EDISON Project, Editors Juan J. Cuadrado-Gallego, Yuri Demchenko, Springer Nature Switzerland AG 2020, ISBN 978-3-030-51022-0, ISBN 978-3-030-51023-7
[2] Artificial Intelligence, Study Programme, University of Amsterdam [online] https://www.uva.nl/shared-content/programmas/en/masters/artificial-intelligence/study-programme/study-programme.html

Keywords:

Artificial Intelligence, Artificial Intelligence Competences, Body of Knowledge, Artificial Intelligence Educations, Machine Learning, EDISON Data Science Framework (EDSF).