=Paper=
{{Paper
|id=Vol-3191/paper25
|storemode=property
|title=Towards Building an AI Curriculum for High School Students (short paper)
|pdfUrl=https://ceur-ws.org/Vol-3191/paper25.pdf
|volume=Vol-3191
|authors=Maria Nisheva-Pavlova
}}
==Towards Building an AI Curriculum for High School Students (short paper)==
https://ceur-ws.org/Vol-3191/paper25.pdf
Towards Building an AI Curriculum for High School
Students
Maria Nisheva-Pavlova 1, 2
1
Faculty of Mathematics and Informatics – Sofia University St. Kliment Ohridski, 5
James Bourchier Blvd., Sofia, 1164, Bulgaria
2
Institute of Mathematics and Informatics – Bulgarian Academy of Sciences, 8 Acad.
Georgi Bonchev Str., Sofia, 1113, Bulgaria
Abstract
The paper discusses some issues in connection with the design of appropriate
high school curricula, related to artificial intelligence. Further development
and concretization of some previous results of the author on this topic are
presented. A model curriculum of an artificial intelligence course has been
proposed, which can play the role of an elective module for specialized
training in informatics in high school.
Keywords
Artificial intelligence, high school education, problem solving, knowledge
representation and reasoning, uncertainty, machine learning, neural network
1. Introduction
Over the last decade, there has been a particularly strong interest in AI and all
AI-related activities and initiatives. In particular, the interest in appropriate inclu-
sion of AI topics in education at all levels has significantly increased. And while
there are long-established good practices and standards for higher education in
AI, the issues of the gradual inclusion of AI topics at the level of secondary and
high school education still do not have clear solutions.
The paper is a continuation of the author’s research on the development of
AI courses for secondary and high school [1], proposing a model curriculum for
an Artificial Intelligence course for high school students.
2. Background and related work
This section discusses a number of strategic visions and common guidelines
for the process of including various aspects of AI in school curricula. Some good
Information Systems & Grid Technologies: Fifteenth International Conference ISGT’2022, May 27–28, 2022, Sofia, Bulgaria
EMAIL: marian@fmi.uni-sofia.bg (M. Nisheva-Pavlova)
ORCID: 0000-0002-9917-9535 (M. Nisheva-Pavlova)
© 2022 Copyright for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR Workshop Proceedings (CEUR-WS.org)
practices of existing initiatives for design of AI courses, suitable for secondary
and high school students, have been presented as well.
2.1. Strategic frameworks
The most comprehensive vision of the professional AI community for the
process of including various aspects of AI in school and university curricula is
the Beijing Consensus on Artificial Intelligence and Education [2] which sets out
a number of common guidelines for this process. The most significant of them are
based on some common principles such as: “take institutional actions to enhance
AI literacy across all layers of society”; “develop local AI talent, in order to cre-
ate a massive pool of local AI professionals who have the expertise to design,
program and develop AI systems”; “be mindful of the importance of adopting
principles of ethics-, privacy- and security-by-design”; “support the integration
of AI skills into ICT competency frameworks”.
The strategic documents at European and national level (e.g. [3], [4], and [5])
provide specific measures, the implementation of which would allow the educa-
tion system in Bulgaria to develop the knowledge and skills needed for work in
the field and AI, as well as for work in an AI environment. For the level of sec-
ondary and high school education, these measures are formulated quite generally
and include [4, 5]:
• significant increase in the role of the so-called STEM (Science, Technol-
ogy, Engineering, and Mathematics) disciplines and the disciplines related to
the acquisition of digital competencies in school education;
• acquisition of digital skills specific to the creation and application of AI –
both analytical and applied;
• increasing the competencies and culture of students in the field of ethical
and legal issues related to the use of information technologies;
• focusing school education on the acquisition of four categories of skills
and abilities: cross-sectoral cognitive skills; creative abilities; social and sit-
uational skills; precise abilities related to perception and handling.
The currently established Bulgarian state educational standards and curricula
for specialized training in mathematics, informatics and information technology
in secondary education2 provide a good basis for further activities to implement
the indicated measures.
2.2. Best practices
The largest and most popular program for inclusion of AI in school education
is the AI for K-12 initiative (AI4K12)3 which is aimed at developing guidelines
2
https://mon.bg/upload/24016/ndrb-PP-izm092020.pdf
3
https://ai4k12.org
277
for integration of AI in 12-year school education. The AI4K12 guidelines are or-
ganized in so-called grade band progression charts that include separate K-2, 3-5,
6-8, and 9-12 grade bands. The educational content is still under development.
It is organized in thematic units, grouped around “the five big ideas of AI” (see
Figure 1):
Figure 1: The “Five Big Ideas of AI” (adapted from [6])
• Perception (“Making computers “see” and “hear” well enough for practi-
cal use is one of the most significant achievements of AI to date”),
• Representation and reasoning (“Agents maintain representations of the
world and use them for reasoning. Representation is one of the fundamental
problems of intelligence, both natural and artificial”),
• Learning (“Computers can learn from data. … Many areas of AI have
progressed significantly in recent years thanks to learning algorithms that
create new representations”),
• Natural interaction (“Intelligent agents require many kinds of knowledge
to interact naturally with humans. Agents must be able to converse in human
languages, recognize facial expressions and emotions, and … infer inten-
tions from observed behavior”),
• Societal impact (“AI can impact society in both positive and negative
ways”).
The educational resources developed and/or available under the AI4K12 ini-
tiative4 include a wide variety of books (see Figure 2), book chapters, curriculum
materials, videos, online courses, tutorials, demos, software packages, competi-
tions.
4
https://ai4k12.org/resources/list-of-resources
278
Figure 2: Educational resources available under the AI4K12 initiative
Along with AI courses specifically designed for secondary and high school,
there exist more than 200 massive open online courses (MOOCs) in AI and re-
lated areas and some of them have content and volume, suitable for high school
students. One of the best MOOCs in the world is the Finnish ‘Elements of AI’
course5, whose basic version in English consists of two parts: ‘Introduction to
AI’6 and ‘Building AI’7.
The presentation of the first part is oriented to users who have no mathemati-
cal or computing background and is illustrated by examples in relevant areas of
interest to people of different ages and professions. It consists of six chapters,
each of them divided into three sections: What is AI (How should we define AI,
Related fields, Philosophy of AI); AI problem solving (Search and problem solv-
ing, Solving problems with AI, Search and games); Real world AI (Odds and
probability, The Bayes rule, Naïve Bayes classification); Machine learning (The
types of machine learning, The nearest neighbor classifier, Regression); Neural
networks (Neural network basics, How neural networks are built, Advanced neu-
ral network techniques); Implications (About predicting the future, The societal
implications of AI, Summary).
The second part of the ‘Elements of AI’ course – ‘Building AI’, addresses the
same topics as ‘Introduction to AI’, but is oriented to a more technically compe-
5
https://www.elementsofai.com
6
https://course.elementsofai.com
7
https://buildingai.elementsofai.com
279
tent audience and allows to go into more depth in the basic methods of AI. As we
concluded in [1], “it is suitable for high school students with a mathematics or
natural sciences or technology profile, but its content is available to most students
aged at least 15–16”.
3. Model curriculum project
According to the conclusions and recommendations we formulated in our
previous work [1], the model curriculum is based on an appropriate combination
of the two parts of the ‘Elements of AI’ course, extended with a unit on Knowl-
edge representation and reasoning. Some additional topics in the areas of Deal-
ing with uncertain data and knowledge and Machine learning are also included.
The areas of Planning, Natural language processing, Perception and Robotics
remained uncovered due to fears of excessive volume of the curriculum, but it
is appropriate that they be indirectly addressed in conducting specific trainings
with well-thought-out illustrative examples and topics for homework and small
projects.
Figure 3 shows the structure of the curriculum and the main topics covered
by it. The learning content is structured in seven units: Introduction; AI and prob-
lem solving; Knowledge representation and reasoning; Dealing with uncertain
data and knowledge; Machine learning; Neural networks; Implications and future
of AI.
280
Part 1. Introduction: What is AI?
1.1. How should we define AI?
1.2. Related fields
1.3. Philosophy of AI
Part 2. AI and problem solving
2.1. Search and planning
Solving problems by searching; uninformed search; planning in AI
2.2. Heuristic search
Heuristic functions; informed search strategies; greedy best-first search; A* search; hill
climbing
2.3. Game playing
The minimax algorithm; alpha-beta pruning
Part 3. Knowledge representation and reasoning
3.1. Key concepts. Knowledge engineering
Key terminology; knowledge representation and reasoning; building a knowledge base;
properties of good knowledge bases
3.2. Rule-based knowledge representation and reasoning
Production rule systems; type of reasoning in rule-based systems; rule-based expert systems
3.3. Object-oriented knowledge representation and reasoning
Objects and frames; a basic frame formalism; example of using frames
Part 4. Dealing with uncertain data and knowledge
4.1. Probability for knowledge-based systems
Probability fundamentals; Monte Carlo method; use of conditional probability in making
inferences
4.2. The Bayes rule in knowledge-based systems
The Bayes rule; real-world examples of application of the Bayes rule for probability cal-
culation
4.3. Naïve Bayes classifiers
What does a Naïve Bayes classifier do; real-world examples of application of Naïve Bayes
classifiers
Part 5. Machine learning
5.1. Definition and types of machine learning
Key terminology; supervised learning; unsupervised learning; reinforcement learning
5.2. Classification. NN and k-NN classification
Terminology; vector distances; nearest neighbor; k nearest neighbors
5.3. Regression. Linear regression
Terminology; definition and examples of linear regression; use of linear regression to make
predictions
5.3. Text classification
Basics of working with text via NLP; bags of words; TF-IDF
5.4. Overfitting
Key terminology; risks of overfitting
5.5. Clustering. The k-means method
Terminology; centroid-based clustering and the k-means method; hierarchical clustering
281
Part 6. Neural networks
6.1. Neural network basics. Popular neural network models
Notation; simple computing elements; network structures; perceptrons
6.2. Logistic regression. From logistic regression to neural networks
Terminology; difference between linear regression and logistic regression; sigmoid func-
tions and logistic regression
6.3. Deep learning
Multilayer feed-forward networks; convolutional neural networks; recurrent neural net-
works and transformers
Part 7. Implications and future of AI
7.1. Modern AI. The societal implications of AI
7.2. The hype cycle for AI. Main trends driving near-term AI innovation
7.3. Legal and ethical aspects of AI
Figure 3: Model curriculum – main topics covered
The planning and conducting of Bulgarian language courses based on this
model curriculum requires the creation of a lot of teaching and training materials
in Bulgarian, consistent with the age, background knowledge and life experience
of students. In our opinion, these materials can be based on a proper adaptation
of the Bulgarian version of the ‘Elements of AI’ course in combination with some
textbooks in Bulgarian, issued at different times (such as [7] and [8]). Reduced
and properly simplified Bulgarian versions of Part II and Part III of [9], enriched
with appropriate examples, could be developed and used for the purpose. Some
of the publicly available resources created and/or accumulated within the AI4K12
initiative (such as the online courses and demos for K-12 students as well as the
handouts for AI projects in the classroom [10]) could also be adapted and used.
4. Conclusion
The analysis of the proposed model curriculum shows that it would be com-
prehensible and suitable in scope and content in terms of its applicability for vari-
ous educational purposes, including as an elective module for specialized training
in Informatics in high schools. It corresponds sufficiently to the “Five Big Ideas
of AI” and allows easy expansion in order to cover other areas of AI, in particular
the only relatively underrepresented “big idea” of perception.
Appropriate support with teaching materials in Bulgarian and illustrative ex-
amples suitable for the age, life experience and interests of high school students
would make it applicable for various purposes in Bulgarian schools, as well as in
extracurricular activities with distinguished students.
282
5. Acknowledgements
This research is supported by Project BG05M20P001-1.002-0011 “Centre
of Competence MIRACLe – Mechatronics, Innovation, Robotics, Automation,
Clean Technologies” financed by Operational program “Science and Education
for Smart Growth”, co-financed by the European Regional Development Fund, as
part of the educational program of the Intelligent Urban Environment Lab under
this project.
6. References
[1] M. Nisheva-Pavlova, “AI Courses for Secondary and High School – Com-
parative Analysis and Conclusions”. CEUR Workshop Proceedings, ISSN
1613-0073, Vol. 3061, 2021, pp. 9–16.
[2] Beijing Consensus on Artificial Intelligence and Education: Outcome docu-
ment of the International Conference on Artificial Intelligence and Educa-
tion “Planning Education in the AI Era: Lead the Leap” (16–18 May 2019,
Beijing, China). UNESCOC Digital Library, 2019. URL: https://unesdoc.
unesco.org/ark:/48223/pf0000368303 (last visit on 31.03.2022).
[3] Artificial Intelligence for Europe, SWD(2018) 237 final, F2 –
26.06.2018. URL: https://ec.europa.eu/transparency/documents-register/
detail?ref=COM(2018)237&lang=en (last visit on 31.03.2022).
[4] Ministry of Education and Science, AI in Education and Science: Ideas for
the development and use of AI in education and science in the Republic
of Bulgaria, July 2020 (In Bulgarian: Министерство на образованието и
науката, ИИ в образованието и науката: Идеи за развитието и използ-
ването на ИИ в образованието и науката в Република България, юли
2020). URL: https://www.mon.bg/upload/23352/MON+AI+Doc.pdf (last
visit on 31.03.2022).
[5] MTITS, Concept for the Development of AI in Bulgaria until 2030 (in Bul-
garian: МТИТС, Концепция за развитието на ИИ в България до 2030 г.,
приета от Министерския съвет на 16.12.2020 г.). URL: https://www.
mtitc.government.bg/bg/category/157/koncepciya-za-razvitieto-na-izkust-
veniya-intelekt-v-bulgariya-do-2030-g (last visit on 31.03.2022).
[6] D. Touretzky, C. Gardner-Mccune, “Artificial Intelligence Thinking in
K-12” (book chapter). MIT Press, 2022 (in print). URL: https://ai4k12.
org/wp-content/uploads/2021/08/Touretzky_Gardner-McCune_AI-Think-
ing_2021.pdf (last visit on 31.03.2022).
[7] M. Nisheva, D. Shishkov, Artificial Intelligence. Integral, Dobrich, 1995
(in Bulgarian: М. Нишева, Д. Шишков. Изкуствен интелект. Издател-
ство „Интеграл“, Добрич, 1995).
283
[8] S. Stoyanov, T. Glushkova, J. Todorov, Artificial Intelligence: Problem
Solving by Search. Izkustva, Sofia, 2021 (in Bulgarian: С. Стоянов,
Т. Глушкова, Й. Тодоров, Изкуствен интелект: Решаване на проблеми
чрез търсене. Издателство „Изкуства“, София, 2021).
[9] S. Russell, P. Norvig, Artificial Intelligence: A Modern Approach (3rd ed.).
Pearson Education Ltd., 2010.
[10] D. Touretzky, Hands on AI Projects for the Classroom (book series). ISTE
and General Motors, 2021. URL: https://github.com/touretzkyds/ai4k12/
wiki/Book-Series%3A-Hands-On-AI-Projects-for-the-Classroom (last vis-
it on 31.03.2022).
284