=Paper=
{{Paper
|id=Vol-3691/paper42
|storemode=property
|title=Integrating Machine Learning for the Continuing Education of Science Teachers
|pdfUrl=https://ceur-ws.org/Vol-3691/paper42.pdf
|volume=Vol-3691
|authors=Jhon Alé
|dblpUrl=https://dblp.org/rec/conf/cisetc/Ale23
}}
==Integrating Machine Learning for the Continuing Education of Science Teachers==
https://ceur-ws.org/Vol-3691/paper42.pdf
Integrating Machine Learning for the Continuing
Education of Science Teachers
Jhon Alé1
1 University of Chile, Capitán Ignacio Carrera Pinto 1045, Santiago, Chile
Abstract
This study investigates the perception of 42 science teachers in Chile after participating in a two-week
workshop focused on the curricular integration of Machine Learning technology to enrich their teaching
strategies in the Science course. Using KPSI-type Likert surveys, a pre and posttest was administered to
assess changes in perception, followed by statistical analysis.
The results highlight significant improvements in teachers' perception in key areas, such as digital
citizenship knowledge, digital resource selection to support their teaching, and more positive attitudes
towards the integration of Machine Learning in the classroom. However, significant challenges were
identified related to the conceptualization and application of Machine Learning in the educational
environment.
This study underscores the need to provide additional support and specific training to overcome
barriers to the successful adoption of these technologies in science education. These findings are
relevant for the development of effective teacher training strategies and the promotion of successful
integration of Machine Learning in educational settings.
Keywords
Continuing Education, ICT into the Curriculum, Machine Learning, Science Education, STEM
Education.1
1. Introduction
The potential of technology to strengthen educational systems and advance towards the
achievement of the Sustainable Development Goals is internationally recognized [1]. These
technologies are positioned as strategic factors that contribute to equitable growth and address
the challenges of the 21st century [2]. Pedagogical practice is increasingly influenced by emerging
technologies, such as Artificial Intelligence (AI), Machine Learning (ML), or Extended Realities
(virtual, augmented, or mixed), which pose ethical challenges [3] and essential questions about
how to interact with these technologies, select them, and harness their potential to support
students' teaching and learning, as well as prepare them for the future [4].
Despite the widely accepted importance of teacher knowledge [5, 6] as the cornerstone to
address students' digital competence [7, 8], the reality shows that technology is barely integrated
into classrooms, and when it is, it often follows traditional teaching methods [9]. This situation is
attributed, in part, to the lack of an educational focus in technology research [10], the absence of
new theories, models, and methods for integration into pedagogical practice [11], and the lack of
evidence, especially in the Latin America and the Caribbean region [12].
In this context, the present study focuses on contributing to this purpose and presents the
results of the implementation of a remote workshop for science teachers throughout Chile. The
aim of the workshop is to assess teachers' perceptions of the use of Machine Learning tools in
their science classes.
CISETC 2023: International Congress on Education and Technology in Sciences, December 04–06, 2023, Zacatecas,
Mexico
jhon.ale@ug.uchile.cl (J. Alé-Silva)
0000-0002-1999-4012 (J. Alé-Silva)
© 2023 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)
CEUR
ht
tp:
//
ceur
-ws
.or
g
Works
hop I
SSN1613-
0073
Pr
oceedi
ngs
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
2. Method
The work is oriented towards research in educational design [13]. Workshops and educational
resources are constructed, considering distance work and progressive activities that allow the
curricular integration of Machine Learning technological resources to support the curricular
learning of science.
The teachers participating in the workshop are 22 women and 20 men (n = 42). Moreover, the
majority of them are teachers with more than 5 years of teaching experience in different cities in
Chile.
A 4-point Likert self-assessment survey was used (see Table 1), following the model of the
Knowledge and Prior Study Inventory (KPSI) [14].
Table 1
KPSI Questionnaire questions (Spanish version)
Questions 1 2 3 4
Q1. ¿Puedo explicar el concepto de Machine Learning?
Q2. ¿Conozco formas de incorporar el Machine Learning en mi
ejercicio docente?
Q3. ¿Puedo establecer criterios para identificar y seleccionar
recursos de Machine Learning apropiados para mi ejercicio
docente?
Q4. ¿Soy consciente de los procesos necesarios para integrar
curricularmente recursos de Machine Learning?
Q5. ¿Sé cómo diseñar actividades que integren tecnología de
Machine Learning y sean coherentes con los objetivos de
aprendizaje de ciencias?
Where 1 = Completely Agree, 2 = Agree, 3 = Disagree, 4 = Completely Disagree.
The survey was conducted using a pre-post test to assess changes in the perception of the 42
science teachers. A descriptive statistical analysis was performed to compare significant changes
in the average scores of the teachers' perceptions by comparing the initial results with the final
ones. The responses were coded according to their tendency, with 1 = " Completely Agree or
Agree" and 0 = "Disagree or Completely Disagree”. Non-parametric chi-squared tests and the
McNemar Test for paired samples were utilized in the analysis [15].
2.1. Science Workshop with Machine Learning
The objective of the workshop is to explore and evaluate some applications of Machine Learning
for Science Education, through the creation of various classifiers and decision trees that support
the development of school scientific research and the understanding of the basic processes
underlying Artificial Intelligence. The workshop is primarily aimed at in-service or pre-service
Natural Science teachers in primary or secondary education (Physics, Chemistry, or Biology),
without excluding educators from lower levels. At the end of the workshop, teachers and
educators create simple classifiers (with text, image, audio, or video) using Google's "teachable
machine" software. Teachers use decision trees to test scientific research hypotheses in various
areas and to support the resolution of socio-scientific problems.
At the beginning of the workshop, teachers were given explanations of fundamental concepts
of artificial intelligence and machine learning, accompanied by examples illustrating their
application in the execution of projects in the field of natural sciences. This introductory phase
aimed to establish a common understanding and familiarize participants with the essential
concepts of these technologies in the context of their discipline. This pedagogical strategy helped
ensure that teachers had a solid initial understanding of artificial intelligence and machine
learning, setting the stage for the workshop with a shared knowledge foundation.
During the workshop, an activity involving the construction of decision trees took place. In this
phase, science teachers explored 'training' situations, similar to the machine learning training
process. In these situations, teachers worked with images and undertook the task of
distinguishing between poisonous and non-poisonous animals based on their physical
characteristics, as exemplified in Figure 1. This activity aims for teachers to train and build their
own machine learning model based on a dataset with pre-defined categories, specifically using a
supervised learning model.
During this activity, teachers played an active role in creating decision rules, establishing
criteria and guidelines for making these distinctions. Subsequently, they compared these rules in
'model tests' to evaluate their effectiveness, and they also propose scientific research hypotheses
based on the characteristics described in their decision tree models. This hands-on experience
not only allowed teachers to understand the principles underlying machine learning but also to
apply them concretely and interactively in the classification of animals, enriching their
understanding and skills in the subject.
Figure 1: Activity of Classifying Poisonous Animals (in Spanish). Based on ReadyAI activities
At the end of the experience, teachers share ideas for school scientific projects that integrate
Machine Learning and Artificial Intelligence into the curriculum to support teaching, as shown in
Figure 2.
Figure 2: Brainstorming on Machine Learning in Science Education (Spanish version)
3. Results
The results from Table 2 indicate changes across all average scores of science teachers' responses.
Trends show favorable shifts, notably leaning towards "Agree" or "Strongly Agree" options on the
Likert Scale. The most significant change was observed in question 1 (Q1) concerning perceptions
regarding the ability to explain the concept of Machine Learning.
Table 2
KPSI descriptive statistical results
Questions Mean SD
KPSI Pre-test
Q1 2,83 ,621
Q2 3,14 ,683
Q3 2,45 ,889
Q4 2,52 ,943
Q5 2,26 ,828
KPSI Post-test
Q1 1,64 ,656
Q2 2,19 ,994
Q3 1,86 1,026
Q4 1,98 ,869
Q5 1,64 ,656
As evident in Table 3, when comparing responses in the pre-post KPSI test, significant changes
were revealed (with an asymptotic bilateral significance of p < 0.05) in indicators related to:
• Acquiring knowledge about the integration of Machine Learning in my teaching practice.
• Establishing criteria to identify and select appropriate Machine Learning resources for
my teaching.
• Recognizing the necessary processes for the curricular integration of Machine Learning
resources.
Table 3
Pre-post test significance results
Questions Value gl Asymptotic Bilateral Significance
Q2 28,273 5 <,001
Q3 25,000 3 <,001
Q4 17,000 4 ,004
In the context of this research, it is essential to highlight that the self-assessment of the ability
to 'explain the concept of Machine Learning' did not reveal significant changes in science teachers'
perceptions after completing the workshop. This suggests that, despite having formative
explanations on the topic during the workshops, teachers' ability to communicate and
understand this specific concept requires more focused work to enhance their comprehension.
This finding has important implications for teacher training in the context of science education.
Despite the growing relevance of artificial intelligence and Machine Learning in education, it
seems that teachers do not feel equipped to discuss the subject or explain basic concepts of how
it works.
Similarly, the self-assessment of the ability to design activities that integrate Machine Learning
technology coherently with the curricular objectives of natural sciences did not show significant
differences in teachers' perceptions. This indicates that, although they were given the
opportunity to explore the integration of Machine Learning into their teaching, teachers' ability
to align these activities with the specific objectives of the natural sciences curriculum did not
experience noticeable improvements. These results also suggest the need for more specific
attention in teacher training regarding understanding and the ability to use Machine Learning
technology effectively, as well as the integration of this technology into the natural sciences
curriculum.
In general, it can be inferred that the two-week workshop might not be sufficient to achieve
substantial changes in these areas of teaching competence, as they require more in-depth
attention. These findings emphasize the importance of having more training activities in the
future to develop learning that allows teachers to delve into the understanding and application
of Machine Learning to support education.
4. Conclusion
This research focused on evaluating the perception of 42 science teachers in Chile after
participating in a workshop designed to explore and assess the application of Machine Learning
tools in their science classes. Despite international recognition of the potential of technology in
education, there has been a lack of effective integration of these tools in the classroom. The
research results indicate that, although the workshop provided teachers with an initial
understanding of Machine Learning applications in science, no significant changes were observed
in teachers' ability to explain the concept of Machine Learning or their ability to design activities
that integrate this technology coherently with the curricular objectives of natural sciences. These
findings underscore the need for more comprehensive and long-term training in the field of
emerging technologies in science education.
The workshop proved to be a valuable experience for teachers by allowing them to apply
Machine Learning concepts in practical situations, such as the construction of decision trees.
However, the complexity of understanding and conveying these concepts, as well as effective
integration into the curriculum, requires a deeper and ongoing approach.
Ultimately, this research highlights the importance of continuing to develop training
opportunities for teachers to gain a more solid understanding of artificial intelligence and
Machine Learning, allowing them to fully leverage its potential in science education. Future work
should focus on addressing these shortcomings and enriching teacher training in this critical field
for 21st-century education. To achieve this goal, teachers require more focused and detailed
initial training on the concept of Machine Learning through specific pedagogical strategies. It is
crucial to advance the effective integration of Machine Learning into the curriculum, enabling
teachers to design activities that align with subjects and connect with real-world situations such
as disease diagnosis prediction or machine diagnostics using Machine Learning [16, 17]. This
involves developing didactic resources and teaching strategies that link technology with
curriculum content. Furthermore, for the future, it is proposed to implement and evaluate
extended training courses for teachers, focusing on the use and integration of Artificial
Intelligence in scientific education, tailored to their specific needs and contexts.
Acknowledgements
This work has been developed with the support of the University of Chile.
References
[1] M. Fengchun, H. Wayne, Guidance for generative AI in education and research, UNESCO
(2023). https://unesdoc.unesco.org/ark:/48223/pf0000386693
[2] UNESCO, The International Science and Evidence based Education (ISEE) Assessment: 2.6
Education Technology, UNESCO (2023). https://mgiep.unesco.org/iseeareport
[3] S. Giannini, Generative AI and the future of education, UNESCO (2023).
https://unesdoc.unesco.org/ark:/48223/pf0000385877
[4] K. Zhang, A. Begum Aslan, AI technologies for education: Recent research & future directions,
Computers and Education: Artificial Intelligence 2, (2021).
https://doi.org/10.1016/j.caeai.2021.100025
[5] P. Mishra, M.J. Koehler, Technological Pedagogical Content Knowledge: A Framework for
Teacher Knowledge. Teachers College Record (2006), 1017–1054.
https://doi.org/10.1111/j.1467-9620.2006.00684.x
[6] UNESCO, UNESCO ICT Competency Framework for Teachers, UNESCO (2018),
https://unesdoc.unesco.org/ark:/48223/pf0000265721
[7] L. Cuban, H. Kirkpatrick, C. Peck, High Access and Low Use of Technologies in High School
Classrooms: Explaining an Apparent Paradox. American Educational Research Journal,
(2001), 813-834. https://doi.org/10.3102/0002831203800481
[8] OECD, OECD FUTURE OF EDUCATION AND SKILLS 2030 OECD Learning Compass 2030 A
SERIES OF CONCEPT NOTES, OECD (2020). https://www.oecd.org/education/2030-
project/teaching-and-learning/learning/learning-compass-
2030/OECD_Learning_Compass_2030_Concept_Note_Series.pdf
[9] L. Ilomäki, M. Lakkala, Digital technology and practices for school improvement: innovative
digital school model. RPTEL 13, 25 (2018). https://doi.org/10.1186/s41039-018-0094-8
[10] X. Chen, H. Xie, D. Zou, G.J. Hwang Application and theory gaps during the rise of artificial
intelligence in education Computers and Education: Artificial Intelligence, 1 (2020), p.
100002. https://doi.org/10.1016/j.caeai.2020.100002
[11] K.Y. Tang, C.Y. Chang, G.J. Hwang Trends in artificial intelligence-supported e-learning: A
systematic review and co-citation network analysis (1998–2019) Interactive learning
environments (2021), 1-19. https://doi.org/10.1080/10494820.2021.1875001
[12] UNESCO, Global education monitoring report, 2023: technology in education: a tool on whose
terms?, UNESCO. https://doi.org/10.54676/UZQV8501
[13] T. Plomp, N. Nieveen, An Introduction to Educational Design Research. Proceedings of the
seminar conducted at the East China Normal University, (2007).
https://ris.utwente.nl/ws/portalfiles/portal/14472302/Introduction_20to_20education_2
0design_20research.pdf
[14] D. Young, P. Tamir, Finding out what Students know. The Science Teacher, 44, 27- 28 (1977).
[15] JB du Prel, B Röhrig, G Hommel, M Blettner. Choosing statistical tests: part 12 of a series on
evaluation of scientific publications. Dtsch Arztebl Int. (2010), 8.
[16] R. Araya, M. Isoda, J. van der Molen Moris. Developing Computational Thinking Teaching
Strategies to Model Pandemics and Containment Measures. Int J Environ Res Public Health,
18 (2021), 12520. https://doi.org/10.3390/ijerph182312520
[17] R. Araya. What and How to Teach Mathematics for the Future?. Association of Mathematics
Educators (2023), 84-108. https://ame.org.sg/2023/09/06/tme2023-vol-4-no-2-pp-84-
108/