Professional Documents
Culture Documents
BIOLA UNIVERSITY
by
W
has been approved by the
IE School of Education
Fall 2020
PR
W
IE
EV
Table of Contents
Abstract…………………………………………………………………………………………... 1
Chapter/Section
W
Create an Adaptive Learning Environment …………………………………………… 11
IE
Provide Collaborative Learning Experience …………………………………………… 14
Participant ……………………………………………………………………………………… 27
Discussion …………………………………………………………………………………… 43
W
Conclusion …………………………………………………………………………………… 47
IE
Limitations …………………………………………………………………………………… 48
Abstract
The potential and benefit of artificial intelligence (AI) in many aspects have been explored
tutoring system, automated scoring system, and predictive model are believed to be able to
enhance the entire learning process or even create a new type of learning pattern. Putting ethical
considerations and the drawbacks of artificial intelligence aside, researchers and software
developers believe that the popularity of AIEd will come only sooner rather than later. Although
the view of teachers is critical of policymakers, researchers, or even software developers who
support this technology, the perspectives of public elementary school teachers are less discussed
W
in literature. Therefore, a qualitative phenomenological research design was used to investigate
IE
the perspectives of public elementary school teachers on AIEd. The findings of the study showed
that most teachers were willing to integrate AIEd in their teaching profession when opportunities
EV
were granted. More research reporting the effectiveness of AIEd might help in increasing the
reduced human interaction, the biggest factors that inhibited teachers from integrating AIEd.
PR
This point of view was particularly mentioned from their experience of providing distance
learning instruction in the midst of the COVID-19 pandemic. The findings of the study also
revealed that teachers believe that technology is a supportive mechanism aiding teachers in
Chapter 1: Introduction
has also started experiencing the ripple effects of AI and AI’s potential has received academic
attention since five decades ago (Self, 2016). Today, considering the advancement of the 5G
network, the technology of AI sped up simultaneously given the higher capacity of supply
toward infrastructures and massive amounts of data that AI needs. Moreover, the recent virus
W
outbreak of COVID-19 inexorably facilitated a learning system that enables long-distance and
IE
data-sharing features in which AI technology might enhance the education system. The fact that
an increased number of AI researchers, venture capitalists, and product developers are shifting
EV
their attention to education is exciting yet concerning because this shift may drastically shape the
unfolding. Thus, different types of research play a significant role in helping educators to prepare
PR
the students for new roles that the economy will create.
Before predicting what kind of new education era will be created, it is important to
explore how the current projects and products address the challenges faced by today’s education
system. This paper gathered the general meanings of AI and artificial intelligence in education
(AIEd) with a combination of several versions of meanings. Also, this paper discussed how AIEd
helps to solve the major challenges that most teachers and schools struggle with nowadays. For
example, the challenges include increasing the availability to create individualized instruction to
accommodate students’ specific needs and offering more time for teachers to prepare better
PERSPECTIVES ON AIED 3
interventions rather than devoting most of the time handling administrative work (Baker &
Smith, 2019). In this paper, the goals of AIEd were to provide an adaptive learning environment,
promote a collaborative learning experience, and help to reduce administrative burden using
prediction models. Although the opportunities of AIEd were discussed, the challenges of
implementing AIEd needed to be considered as well. The challenges included the lack of
understanding of algorithms and ethical issues involving safety and bias. Furthermore, the paper
presented several possible futures that could be created by AIEd. Whether a utopian view or a
dystopian view is adopted, the fact that many countries have involved AIEd in their national
blueprints provokes the need for educators to study the topic in-depth (Baker & Smith, 2019;
W
Pedro, 2019).
IE
Getting to know someone’s perspective is crucial in understanding a person’s experience,
situation, and the tenor of their attitudes (Maibom, 2019). In this case, teachers’ perspectives on
EV
AIEd are important for many future undertakings in terms of the strategy of promoting AIEd and
the improvisation of AIEd. Although researchers have reported the necessity and benefit of
integrating AI in education and even discussed the possible trends of AIEd in the future, the
PR
most research was conducted in a higher education setting rather than the elementary level
(Hinojo-Lucena et al., 2019; Khare et al., 2018; Popenici & Kerr, 2017; Thompson et al., 2018;
Zawacki-Ritchter et al., 2019). This lacuna, therefore, drew attention to the need to investigate
the topic by collecting elementary school teachers’ perspectives. Moreover, amidst the
might be influenced by the experience of social distancing and even distance learning. Therefore,
using a qualitative phenomenological research design, this study aimed to understand the
PERSPECTIVES ON AIED 4
perspectives of public elementary school teachers on AIEd, and the following research questions
were specified:
3. What are the roles of AIEd in the teaching profession based on teachers’ understanding?
4. What are teachers’ concerns about implementing AIEd in their teaching profession?
W
IE
EV
PR
PERSPECTIVES ON AIED 5
Artificial Intelligence
The technological term artificial intelligence (AI) is no longer an unheard idea to many.
However, through all the years of development, definitions of AI are inconsistent because of its
constantly shifting and complex interdisciplinary nature (Kaplan & Haenlein, 2019; Luckin et
al., 2016; Renz et al., 2020). According to Bostrom, AI is not easily recognized, even when the
technology is permeated into general applications, when the applications such as a computer, a
system, or a software become useful and common (CNN, 2006). For instance, the wide
W
collective intelligence, computerized vision, common sense, data mining, emotional intelligence,
IE
games theory, image processing, language processing, neural networks, pattern recognition, and
robotics. Moreover, another reason for the difficulty in defining AI is its penetration in multiple
EV
disciplines: for example, anthropology, biology, computer science, education, finance, law,
linguistics, medicine, neuroscience, philosophy, and psychology, where each group offers
different terminology and perspective (Luckin et al., 2016; Popenici & Kerr, 2017; Russell &
PR
Novig, 2010). Nevertheless, to introduce the meaning of AI, this paper collected explanations
In 1956, McCarthy first introduced the term artificial intelligence in the Dartmouth
Summer Research Project in Artificial Intelligence and coined that “the study of AI is to proceed
on the bases of the conjecture that every aspect of learning or any other feature of intelligence
can in principle be so precisely described that a machine can be made to simulate it” (Russell &
Norvig, 2010, p. 17). According to the first handbook of AI, AI is a computer system associating
with human characteristics where the system performs intelligent skills and behaviors such as
PERSPECTIVES ON AIED 6
understanding language, learning, reasoning, and solving problems (Barr & Feigenbaum, 1981).
After a few decades, the recent interpretations of AI converged to the similar concepts presented
by the previous works. Luckin et al. (2016) suggested that AI is a programmed computer
software using algorithms to process the knowledge of the world with required human
algorithms and techniques that allow computers and machines to simulate human perception and
describe a variety of technologies and methods, such as deep learning, natural language
processing, neural networks, machine learning, or any possible algorithms, Baker and Smith
W
(2019) defined AI as “computers which perform cognitive tasks, usually associated with human
IE
minds, particularly learning and problem-solving” (p. 10).
Although the existing literature has presented a plethora of explanations to define AI,
EV
those definitions still direct the readers to reach some common understanding despite its
extendable and constantly shifting nature. First, AI refers to a computer system, which is also
known as a machine that uses algorithms to perform its operations. Second, AI simulates, instead
PR
of creates, human intelligence and behavior. Third, AI targets to complete tasks, including
Despite the multiform of technologies related to AI study, current research and products
do not establish an omnipotent superintelligence that is mostly depicted in fictional series. The
AI that has been massively discussed in literature is narrow AI—intelligence that targets the data
from one specific domain and translates it to producing one specific outcome (Lee, 2018). Only
if the major constraints of narrow AI are removed and AI can access multi-domain learning,
which can be hardly achieved in the near future, is it possible to create artificial general
PERSPECTIVES ON AIED 7
intelligence—technology that can do many tasks a human can (Baker & Smith, 2019). Therefore,
instead of drifting into a powerful realm of artificial general intelligence, this literature review
focused only on the potential applications of narrow AI as parts of the educational system.
Because the first artificial intelligence in education (AIEd) research was conducted in the
1970s, and the first AIEd conference was held in 1983, the continuous advancement of the
technology in the field has caused the topics relating to AIEd to be actively studied by
educational researchers. From then on, the deepened weaving of AI into the fabric of educational
aspects has caught the attention of many educators as the usage of computer-based learning
W
systems have been drastically increased in the K-12 systems as well as higher education. As for
IE
the recent global updates, the unexpected outbreak of COVID-19 dramatically has increased the
integration of remote learning in teachers’ daily basis and further expanded the territory of AIEd
EV
(Rauf, 2020; Wang, 2020). Nevertheless, similar to AI, whether among the tutor, the tool, or the
learner, setting an explicit explanation to define AIEd is difficult (Cumming, 1998; Dillenbourg,
2016). Cumming (1998) suggested that the difficulty is caused by the blurred distinction of the
PR
wide range of the systems acting as the tutor and the tool. Besides, research contributors could
not draw the boundaries between the users and the systems because too much overlapping
In general, Luckin et al. (2016) suggested that AIEd would unwrap the black box of
learning, and further unravel the missing pieces of how learning actually happens. To achieve
this great mission, AIEd ought to become more computational and scientifically combining the
implicit side of educational, psychological, and social knowledge (Self, 1998). More specifically,
integrating the human-like properties of AI into the fields—such as adapting, synthesizing, self-
PERSPECTIVES ON AIED 8
correcting, and the use of data for processing complex tasks—allow AI to be a computing system
that can enhance the learning process (Popenici & Kerr, 2017). Put simply, Zanetti et al. (2020)
defined artificial intelligence in education as “a system’s ability to interpret data inputted by the
teacher or the student, learn from such data and use those learnings to achieve specific teaching
goals and tasks through flexible adaptation to the environment” (p. 94). By other means, AIEd
explores the connection between a system, learner, and pedagogical content, through the data
analysis and algorithm calculation, eventually leading to task completion and goal achievement.
Models of AIEd
The following text introduced several categories and models of AIEd relating to different
W
themes. First, Luckin et al. (2016) categorized AIEd into three categories: the pedagogical
IE
model, the learner model, and the domain model. Each model discusses different aspects of
knowledge: for instance, the pedagogical model refers to the knowledge and expertise of
EV
teaching, the domain model refers to the knowledge of subjects being learned, and the learner
model refers to the knowledge of the student. To understand what happens between the computer
and the learner, the interactions, such as a student’s current activities, prior achievements,
PR
emotional state, and whether or not they responded to the feedback, can be represented in learner
models. These interactions are then used by the domain and pedagogical models, using
algorithms to process that information, in order to evaluate the progress of the learner and the
teacher. The processed data are then used to determine the next most appropriate content to be
delivered to the learner according to their needs and further create new interactions. All three
models get updates and become “smarter” when more information, activities, and feedback are
collected.
PERSPECTIVES ON AIED 9
Second, although some products and tools combine properties from multiple categories,
Baker and Smith (2019) also grouped AIEd tools in three categories: learner-facing, teacher-
facing, and system-facing. Learner-facing AIEd is the most common category when most people
think of AIEd because it includes the software that students used to learn additional information
that is tailored to students’ needs. This category is usually known as “intelligent tutoring systems
learners. Such tools can be used outside the classroom where students familiarize themselves
with new concepts while classroom time is used to develop knowledge of those concepts. On the
W
other hand, with teacher-facing AIEd that supports automation of tasks such as assessment and
IE
plagiarism detection, teachers could easily provide insightful feedback while cooperating with
the learner-facing AIEd data. Additionally, teachers can offer various methods of teaching based
EV
on the shared characteristics of students. For example, teachers can try experimenting with
different small collaborating groups or planning class seating plans that reduce behavioral
problems. Baker and Smith (2019) noted that system-facing AIEd is the least explored subject. In
PR
Instead of classifying AIEd based on how different facets view the capabilities of each
model and system, Murphy (2019) suggested that AIEd can be categorized based on the systemic
design of the applications. The narrow AI that has been applied in most of the AIEd studies fall
into two categories, which are rule-based expert systems and machine-based learning systems.
The first category includes applications that are used to function as adaptive instructional
software systems. Two key elements of expert systems are the knowledge base and the inference
PERSPECTIVES ON AIED 10
engine. The knowledge base collects the encoded data and experience needed for problem
solving, and often follows if-then statements or rules. The inference engine makes decisions
based on incoming sources of data and generates recommendations that abide by the rules in the
expert knowledge base. In contrast to the rule-based expert system, machine-based learning is an
approach to build a predictive model by processing massive amounts of data using statistical
algorithms. The system discovers patterns among the input predictor variables with the output
variable of interests. For example, machine-learning algorithms might be used to analyze the
relationship between student characteristics in their early school years and their on-time
graduation in high school (Barrett et al., 2019; Murphy, 2019). Another promising application of
W
machine learning is the automated essay scoring system (AES). With the technology of natural
IE
language processing, the algorithms extract features of the text and compare them with the
features of human-scored essays. Ultimately, the algorithms learn the patterns of relationships
EV
between the features and different writing levels to grade similarly to human scorers on standard
The categories, which are mentioned by different authors to explain the types of AIEd,
PR
generally link the foundational design of the technology to several educational purposes. Despite
the disparities between the models, AIEd is believed to be an educational tool that facilitates the
learners, teachers, and systems through a continually advancing nature of the technology.
Goals of AIEd
Before delving deeper into the goals of AIEd, having a brief insight of current
educational challenges and future educational circumstances is essential to understand how AIEd
may assist individual learners, group learners, teachers, and school administrators. According to
Baker and Smith (2019), several educational challenges that might be resolved by the
PERSPECTIVES ON AIED 11
implementation of AIEd are the following: (a) teachers occupied with heavy workload, (b) “one-
size-fits-all” instruction that is inflexible, (c) limited assessments for both student and teacher,
(d) the lack of sharing between educational organizations, and (e) educational inequality.
Furthermore, either directly or indirectly extended from these issues, some educators suggested
that future educational settings should (a) provide students with knowledge of new jobs that the
economy will create, (b) teach students more 21st-century skills, and (c) create a lifelong
learning society for all (Kaplan & Haenlein, 2019; Khare et al., 2018; Luckin et al., 2019; Woolf
et al., 2013). For the most part, these challenges and circumstances are largely aligned with the
goals of making learning more diverse, accessible, direct, social, collaborative, related, and
W
available to people (Woolf et al., 2013). By combining the characteristics of AIEd, researchers
IE
and developers believe that AIEd can improve these problems to some extent and construct the
desired future educational setting. For learners themselves, creating an adaptive learning
EV
environment and a collaborative learning experience is the solution that can be offered by AIEd
to address the problems. In terms of administration, researchers have highlighted the predictive
The underlying principle of the adaptive learning system is built on the knowledge of
apprenticeship training and human tutor (Shemshack & Spector, 2020). Today, even though
learning is still mostly used interchangeably with personalized learning, which has a meaning of
instruction, and differentiated instruction. Unlike the early forms of adaptive learning that only
serve a small number of learners, modern adaptive learning aims to deal with large numbers of
Reproduced with permission of copyright owner. Further reproduction prohibited without permission.