Content uploaded by Leif Sundberg
Author content
All content in this area was uploaded by Leif Sundberg on Feb 06, 2024
Content may be subject to copyright.
Journal of
Information
Systems
Education
Volume 35
Issue 1
Winter 2024
Teaching Tip
Using No-Code AI to Teach Machine Learning in Higher
Education
Leif Sundberg and Jonny Holmström
Recommended Citation: Sundberg, L., & Holmström, J. (2024). Teaching Tip:
Using No-Code AI to Teach Machine Learning in Higher Education. Journal of
Information Systems Education, 35(1), 56-66. https://doi.org/10.62273/CYPL2902
Article Link: https://jise.org/Volume35/n1/JISE2024v35n1pp56-66.html
Received: December 5, 2022
First Decision: March 1, 2023
Accepted: May 2, 2023
Published: March 15, 2024
Find archived papers, submission instructions, terms of use, and much more at the JISE website:
https://jise.org
ISSN: 2574-3872 (Online) 1055-3096 (Print)
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
56
Teaching Tip
Using No-Code AI to Teach Machine Learning in Higher
Education
Leif Sundberg
Jonny Holmström
Swedish Center for Digital Innovation, Department of Informatics
Umeå University
Umeå, Sweden
leif.sundberg@umu.se, jonny.holmstrom@umu.se
ABSTRACT
With recent advances in artificial intelligence (AI), machine learning (ML) has been identified as particularly useful for
organizations seeking to create value from data. However, as ML is commonly associated with technical professions, such as
computer science and engineering, incorporating training in the use of ML into non-technical educational programs, such as social
sciences courses, is challenging. Here, we present an approach to address this challenge by using no-code AI in a course for
university students with diverse educational backgrounds. This approach was tested in an empirical, case-based educational setting,
in which students engaged in data collection and trained ML models using a no-code AI platform. In addition, a framework
consisting of five principles of instruction (problem-centered learning, activation, demonstration, application, and integration) was
applied. This paper contributes to the literature on IS education by providing information for instructors on how to incorporate no-
code AI in their courses and insights into the benefits and challenges of using no-code AI tools to support the ML workflow in
educational settings.
Keywords: Artificial intelligence, Machine learning, IS education research, Information systems education
1. INTRODUCTION
Machine learning (ML), a subfield of artificial intelligence
(AI), focuses on the development, application, and analysis of
computer systems capable of learning from experience. In a
common variant, supervised ML, a system is shown numerous
examples of a type of data, e.g., images or texts describing
particular objects or phenomena, to train it to “learn” or
recognize patterns in them. The system can then use this
learning to predict new “unseen” data, i.e., data it has not
previously encountered (Jordan & Mitchell, 2015; Kühl et al.,
2022). Leavitt et al. (2021, p. 750) define ML as “a broad subset
of artificial intelligence, wherein a computer program applies
algorithms and statistical models to construct complex patterns
of inference within data” (see also Bishop, 2006).
Massive increases in the processing power of digital
technology and available data, in combination with better
algorithms, e.g., deep learning algorithms (see Lecun et al.,
2015) have set the stage for increases in the use of ML in many
contexts (Dwivedi et al., 2021). Accordingly, organizations are
increasingly deploying intelligent systems that can process
large amounts of data, provide knowledge and insights, and
operate autonomously (Simsek et al., 2019; Sturm et al., 2021).
As noted by Ma and Siau (2019, p. 1), “Higher education
needs to change and evolve quickly and continuously to prepare
students for the upheavals in the job market caused by AI,
machine learning, and automation.” Among other things, these
authors argue that AI must be integrated into academic
curricula, and not only those of science, technology,
engineering, and mathematics (STEM) departments. However,
despite abundant research on applications of AI in educational
settings (e.g., Humble & Mozelius, 2022; Luan & Tsai, 2021),
much less attention has been paid to instruction of students with
non-technical backgrounds in ML’s practical use and
applications (Kayhan, 2022). As ML is commonly associated
with technical professions, such as computer science and
engineering, incorporating training in its use into non-technical
educational programs, such as business- and management-
oriented social sciences and Information Systems (IS)
programs, is challenging. Similar issues have been raised in
previous research on novel intelligent systems (Liebowitz,
1992, 1995) as educators have sought to integrate their use into
business and IS programs. Recently, scholars have identified a
need to integrate AI curricula in ways that enable students to
develop a sufficient understanding of technology such as ML to
apply it without detailed knowledge of AI algorithms (Chen,
2022). In this paper, we assess “no-code” AI platforms’
potential utility in an effort to meet this need. In contrast to
conventional AI systems, which require significant resources
for installation and use, these platforms can be readily applied
in educational contexts. Thus, they are easy-to-use and
affordable forms of AI, and they guide users through the
process of developing and deploying AI models, with no need
to learn all about the intricacies associated with complex
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
57
algorithms (Lins et al., 2021; Richardson & Ojeda, 2022).
Hence, in this paper, we pose two research questions (RQs):
RQ1: How can no-code AI be used to teach ML in non-
technical educational programs?
RQ2: What are the benefits and challenges of using no-
code AI in education?
As already mentioned, “non-technical” refers here to non-
STEM programs, such as business- and management-oriented
courses. To answer the RQs, we present a teaching tip based on
a case study of a master’s level AI for Business course at Umeå
University, Sweden, in which qualitative data were collected
through interactions with, and observations of, the students. In
the remaining sections of the paper: we summarize previous
research on no-code software, describe the educational setting,
describe the materials and methods used, present the results,
discuss them, and finally offer concluding remarks.
2. BACKGROUND: TOWARDS “LIGHTWEIGHT” AI
In this section, we present a brief overview of the ML workflow
(subsection 2.1) and then summarize the literature on the
emergence of no-code AI platforms (subsection 2.2).
2.1 What is Machine Learning?
ML refers to a broad set of AI applications in which computers
build models based on patterns they recognize in datasets and
use the models to generate hypotheses about the world. Such
models have myriads of uses in problem-solving software
exploited in industrial and other organizations (Russell &
Norvig, 2022). The general ML workflow (e.g., Chapman et al.,
1999; Kelleher & Brendan, 2018; Schröer et al., 2021) begins
with the creation of a training dataset from which a machine can
learn something (Figure 1). Most applications today are based
on supervised learning procedures through which a machine
learns from labeled data, e.g., text describing an image, such as
a photo or drawing of a dog or cat (Fredriksson et al., 2020).
Then the training dataset is processed by an algorithm that
“trains” the machine to recognize corresponding patterns. The
outcome of this process is an ML model that can be used to
make predictions regarding previously unseen data. During the
training process, part of a dataset (e.g., 20% of the images in an
image classifier case) is reserved for testing the model to avoid
problems such as overfitting. Acceptable performance of the
model on the test datasets indicates that it may be used to solve
problems in real-world contexts, such as organizational
settings, if the data provide relevant representations of the
things or phenomena that must be recognized to solve the
problems.
This description is a somewhat simplified version of the
ML workflow. In reality, it takes several iterations of data
collection loops and knowledge consolidation processes to
create a model that provides meaningful results as experts may
have diverging perceptions of what data represent (see Lebovitz
et al., 2021 for a detailed discussion on experts’ disagreements
during data annotation).
2.2 No-Code AI
No-code solutions for software development have been subject
to previous research as they enable non-programmers with little
or no coding experience to produce various applications
(Bhattacharyya & Kumar, 2021; Luo et al., 2021; Lethbridge,
2021; Sahay et al., 2020; Yan, 2021). By adopting low-code
principles, enterprises may not only save time and costs but also
narrow the gaps between business operations and information
technologies, thereby enabling more rapid development and
improvements in product and service quality (Rokis &
Kirikova, 2022).
Figure 1. A Simplified Machine Learning Workflow
As noted by Sundberg and Holmström (2022, 2023), a new
generation of “lightweight” no-code AI platforms—also known
as AI as a service (Lins et al., 2021) or simply AI service (Geske
et al., 2021) platforms—enables non-data scientists to train ML
models to make predictions. Such platforms may match, or even
outperform, coded solutions (Kling et al., 2022). Hence, no-
code AI platforms may be widely applied in diverse settings,
including citizen science, and as low-cost solutions in emerging
markets. In the long run, it has been argued that access to user-
friendly, low-code AI could democratize the adoption of these
systems and stimulate their multidisciplinary use (How et al.,
2021). For example, new “drag-and-drop” interfaces enable
anyone to develop, train, and test AI algorithms in a few hours.
In combination with a range of open-source solutions and
plugins, this vastly simplifies algorithm development and
deployment (Coffin, 2021). The advances are so rapid that
within two years of Woo (2020, p. 961) stating that “AI might
be able to automatically produce code,” advances in generative
AI, tools such as GitHub Copilot and ChatGPT are enabling
code generation based on the input of a user. Computer
scientists have always dreamt of writing programs that write
themselves, and the dream is becoming a commonplace reality.
Recently, academic researchers have also recognized the
powerful potential utility of no-code apps in educational
settings. For instance, Wang and Wang (2021) argue that no-
code (or low-code) app development is transforming traditional
software development practices and present a teaching case
involving the development of a business app.
3. EDUCATIONAL SETTING
As noted by Holmström et al. (2011), rapid technological
developments create challenges for maintaining up-to-date
curricula for educating professionals who will work in
environments with high levels of technology. They highlight
several important issues regarding IS teaching, including the
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
58
importance of ensuring that the students acquire practically
relevant skills through the use of appropriate pedagogical
approaches and generic types of knowledge. As AI is being
increasingly adopted in diverse domains (Dwivedi et al., 2021),
most, if not all, professionals will engage with or be affected by
intelligent systems in their careers. However, as mentioned, AI
is associated with the need to understand algorithms, and hence,
skills rooted in computer science and engineering. This poses
challenges for professionals rooted in other disciplines, not
because they have nothing to contribute to AI or gain from its
use, but because of a lack of fundamental knowledge of how,
for example, an ML system works. A potential remedy, also
already mentioned, is to use “lightweight” AI (Sundberg &
Holmström, 2022) in the form of AI service platforms (Geske
et al., 2021; Lins et al., 2021), which are easy to use with little
to no installation requirements (as they are cloud-based) and
have graphical interfaces that help users to train ML models.
Here we present an approach for using such a system, the
Peltarion (2022) “no-code” deep learning AI platform
(hereafter “the no-code AI platform,” or just “the platform”), in
a higher education setting at the Department of Informatics,
Umeå University, Sweden. The department is part of the
university’s faculty of social sciences and provides three
undergraduate educational programs (on behavioral science
with an orientation towards IT environments, digital media
production, and system science) and two master programs (on
human-computer interaction and IT management), together
with individual courses.
The mentioned AI solution enables non-data scientists to
upload data and then train and evaluate an ML model that can
be deployed via an application programming interface (API).
The platform guides users via a graphical interface together
with suggestions regarding problem types, workflows, pre-
trained models, and iterative improvements. The platform was
used in an “AI for Business” course (15 credits) at Umeå
University, to give the students hands-on experience in training
ML models by engaging in a case-based task. The course is
open for students with diverse educational backgrounds, as
requirements for enrolment are 90 credits in informatics,
computer science, business administration, media and
communication studies, pedagogics, psychology, political
science, sociology (or equivalent competence). In line with the
course curriculum (Umeå University, 2022), the learning
objectives of the exercise were to “Account for and explain the
role of AI in organizational value creation,” by giving the
students first-hand experience of training ML models. The
educational approach is further described in the following
section.
4. MATERIALS AND METHODS
To address the RQs posed in Section 1, we followed a group-
based project approach presented by Mathiassen and Purao
(2002) in the course, inviting the students to engage in the
development of ways of working and participating in
communicative activities regarding “real-life” problems. As
noted by Leidner and Jarvenpaa (1995), such approaches
provide opportunities for students to understand the
“messiness” professionals face in the industry, acknowledging
the social situatedness of these contexts, and that the problems
students will face are “unstructured, ambiguous, and immune to
purely technical solutions” (Holmström et al., 2011, p. 2).
We applied the principles of instruction framework
advocated by Merrill (2007, 2013) in the educational setting.
This incorporates five principles summarized in Table 1:
problem-centered learning, activation, demonstration,
application, and integration. The framework provides an
integrated, multi-strand strategy for teaching students how to
solve real-world problems or complete complex real-world
tasks.
Principle
Description
Problem-
centered
learning
Humans learn better when they solve
problems, so learning is promoted
when learners acquire skills in real-
world contexts.
Activation
Learning is promoted when learners
activate existing knowledge and skills
as foundations for a new skill. An
important step here is to start at the
learner’s level. Activation requires
learning activities that stimulate the
development of mental models and
schemes that can help learners
incorporate new knowledge or skills
into their existing knowledge
framework.
Demonstration
Learning is promoted when learners
observe a demonstration of the skill to
be learned, e.g., by exposure to
examples of good and bad practices.
Application
Learning is promoted when learners
apply new skills they have acquired to
solve problems. Applying new
knowledge or skills to real-world
problems is considered almost
essential for effective learning.
Integration
Learning is promoted when learners
reflect on, discuss, and defend
knowledge or skills they have
acquired. The effectiveness of a course
is enhanced when learners are
provided opportunities to discuss and
reflect on what they have learned in
order to revise, synthesize, recombine,
and modify their new knowledge or
skills.
Table 1. Principles of the Educational Approach
The case presented to the students described a fictive
organization, “WeldCorp,” which specialized in welding,
seeking to expand and acquire customers in additional
geographical markets while retaining and automating quality
measures. To assist the company, we invited the students to
develop ways to use ML as a tool to assess welding points. The
course module described in this paper consisted of a workshop,
a Q&A session, supervising sessions, and a final seminar. Its
content is further outlined in Section 5.1. Nineteen students
attended the course (14 male and five female), with educational
backgrounds including bachelor’s degrees in business and
administration, computer science, and behavioral science. The
empirical materials used in the study presented here, as
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
59
summarized in Table 2, stem from interactions with the
students, the no-code AI platform, and teachers’ reflections.
These materials allowed us to both provide educators with
recommendations for using no-code AI and present interesting
findings on the benefits and challenges associated with these
platforms’ use in educational settings. We identified the
benefits and challenges by subjecting the empirical data to
thematic analysis (Braun & Clarke 2012; Clarke & Braun 2014)
through inductively coding the students’ activities during the
module. More specifically, we coded the activities undertaken
by the students in our empirical setting mentioned and observed
in the materials and then aggregated them into themes,
informed by the steps in the ML workflow presented in
Section 2.1.
Materials
Source(s)
Students’ feedback
and course evaluations
E-mails, notes taken during
the course, written evaluations
and feedback from students.
Students’ written
assignments and
presentations
Two written group reports,
and two presentations during a
final seminar.
Datasets, models and
deployments created
by the students
The Peltarion (2022) no-code
AI platform
Observations
Teachers’ experiences and
reflections during and after
the course
Table 2. Materials and Sources
5. RESULTS: USING NO-CODE AI IN AN
EDUCATIONAL CONTEXT
This section is divided into three parts. In line with Lending and
Vician (2012), in Section 5.1 we provide a description of our
educational procedures to enable instructors to adopt our
approach. Then, the benefits of using no-code AI in education
are presented in Section 5.2, followed by the challenges we
experienced in Section 5.3.
5.1 Detailed Educational Approach
The course module was initiated on December 2, 2021, and the
final seminar was held on January 10, 2022. Thus, the duration
of the module was a little over a month, including Christmas
holiday breaks. The module was initiated with a 3-hour
workshop session that included an introduction to ML, followed
by a demonstration of the no-code AI platform’s functionalities
and a description of the group assignment. The information
presented and considerations applied in this workshop are
summarized in the following text.
As the students came from different backgrounds, it was
clearly stated that the workshop would not include a deep
examination of phenomena such as neural networks and would
focus instead on providing students with sufficient information
to get hands-on experience in training ML or deep learning
(DL) models. An overview of the current status of ML was
presented as increases in the scale of datasets, together with
improvements in algorithms and processing speed have
increased capabilities for machines to “learn.” This included a
presentation of:
• A short video showing how neural networks “see”
things in image data:
https://www.youtube.com/watch?v=xS2G0oolHpo&ab
_channel=NOVAPBSOfficial
• Figures from an overview by Hilbert and López (2011)
of how the capacities of storing data rapidly shifted from
analogue to digital formats.
• A comparison of the world’s fastest supercomputer in
1997 (ASCI Red), which reached a speed of 1.8
teraflops, and the SONY Playstation 3 video game
console that reached the same speed nine years later.
Then, the differences between supervised, unsupervised,
and reinforced ML were briefly presented. We emphasized that
the module would focus largely on supervised learning, the
basis of most commercial and industrial applications of ML
today, so the students would need to engage with data labeling.
This is important for two reasons. First, collecting and
annotating data are crucial but time-consuming activities that
take most of the time spent during ML development
(Fredriksson et al., 2020). Second, if this element is neglected
or poorly done, the resulting ML models will perform poorly
and generate inaccurate, irrelevant, or even harmful results
(Sambasivan et al., 2021).
Next, the lecture outlined the kinds of problems that can be
solved by using ML. As noted by Kayhan (2022, p. 123),
“Many students lack the preparation for the workforce because
they cannot conceptualize valid input-output relationships for
the problems they propose to solve using ML.” Thus, despite
the widespread hype surrounding intelligent systems, there is a
lack of specificity of the kinds of problems algorithms can
actually solve. As noted in Section 2.1, ML is a set of
technologies that involve the training of algorithms to create
models that can provide predictions concerning previously
unseen datasets. Hence, ML cannot solve “general” problems
such as “increasing efficiency” or “improving quality”: they
need specific problem formulations accompanied by relevant
datasets. Thus, in this part of the lecture, we presented a
checklist for determining whether ML would be suitable to
apply:
1. Do you have a use case?
2. Can the use case be solved by AI/ML (or simpler
means)?
3. Do you have data?
4. Do you have annotated data?
We also presented examples of various problems/use cases
that ML can solve, such as anomaly detection, classification
problems (identifying features in texts and images), building
chatbots based on text similarity functions, and various
regression problems, such as predictions of sales and housing
costs. Before demonstrating the functionality of the no-code
platform, we described the ML workflow, both generally as
shown in Figure 1 and more specifically for the Peltarion
platform, as displayed in Figure 2. Although the platform is
now discontinued, this workflow (data collection + preparation,
training, evaluation and deployment of an ML model) is at the
core of most ML development efforts and protocols applied in
other no-code AI platforms (such as BigML, Amazon
SageMaker, Google AutoML and Teachable Machine).
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
60
After presenting the above activities in a traditional lecture,
supplemented by visual aids and other materials, we turned our
attention to the no-code platform.
Figure 2. The ML Workflow in the No-Code AI
Platform
An important step during the use of no-code AI is to check
the requirements of the platform of choice in terms of data types
(e.g., tabular, images, or text). Familiarity with the selected
platform’s tools for processing and labeling data is also
important. Thus, to provide participating students with an
understanding of how the no-code AI platform handled
different data types, we used free datasets from Kaggle (2023):
• To explore tabular data, we used the popular “IRIS”
dataset, which can be used to predict the species of a
flower based on the size of petals and sepals.
• For image data, images of cats and dogs can be used to
train a binary classifier. Images of craters on the Moon
and/or Mars can be used to train object detectors (if this
feature is available in the platform. See Figure 3 for an
example).
• Data from the Internet Movie Database (IMDB) can be
used to predict whether a text is “positive” or “negative”
to train a model that can make predictions based on NLP
(natural language processing).
Figure 3. Image Annotation for Object Detection in the
BigML Platform
During the demonstration of how to upload data, we briefly
described and outlined procedures for various possible formats
for tabular and text data (e.g., CSV and npy), but not procedures
for connecting to “data warehouses” such as BigQuery or Azure
Synapse, as it was irrelevant for the planned task. Instead, we
focused more on how to upload image data to the platform, as
this was the type of data the students would handle in the
following case. An advantage of using no-code AI in such cases
is that images can be annotated by placing them in folders that
act as labels, compressing them into zip files, and then
uploading them to the platform. The platform then takes care of
processing and cropping the images to standardized formats. A
negative effect, which we informed students about, is that
important features near the edges of the images may be cropped.
Then, we demonstrated various examples of ML problems,
and their possible solutions using the no-code AI platform.
Depending on the type of data involved, the platform suggests
certain problems as the user chooses the input (data) and one or
more targets (labels). As mentioned, examples of such
problems include image classifications and image/text
similarity searches. Thus, in this phase, we also displayed
examples of ways to use pre-trained ImageNeT-based and
NLP-based (e.g., BERT) models for classifying and predicting
patterns in images and texts, respectively. The use of pre-
trained models relaxes the requirements to use big datasets, as
users can fine-tune these models with their own data. Links to
online tutorials and datasets (e.g., Kaggle) were uploaded to the
course teaching platform, for students who wanted to proceed
by experimenting with different types of data and problems.
In another important part of this demonstration, we showed
how ML models can be evaluated. This is done by splitting the
dataset(s) into a training set and test (and/or validation) set. The
algorithm is not exposed to the test set during training, so it can
be used to evaluate how a model performs on previously unseen
data. Common pitfalls, such as data bias and overfitting, were
also introduced during this session. The platform enabled the
generation of two indicators that are commonly used for
evaluating models: receiver operating characteristic (ROC)
curves and confusion matrices, which are especially useful for
enhancing students’ understanding of the output of ML models,
and why their deployment requires careful consideration.
Essentially, an image model outputs a probability of what it
thinks is present in an image, e.g., “0.76 cat.” Depending on the
problem at hand, and associated requirements, a threshold can
be set to determine how “certain” a model must be before it can
classify something. Important measures here include accuracy,
recall, and precision. While accuracy is a measure of a model’s
overall performance, there is always a trade-off between recall
and precision. Students can be taught the relevance of this
tradeoff using two types of examples: ML-based spam filters,
and medical diagnostics. When constructing a spam filter, it is
often more important to minimize the number of “false
positives” (potentially important emails that end up in the spam
filter) than the number of “false negatives” (spam emails that
end up in the inbox). Thus, precision is a good measure for such
a model, as it assesses whether what is being classified as
“spam” really is spam. In contrast, during medical diagnosis,
avoiding false negatives is often much more crucial than
avoiding false positives (as assessed by a recall measure),
because wrongly classifying ill people as healthy can have
severe consequences for them. For understanding such issues,
knowledge of ROC curves is important because they illustrate
three key aspects of ML models. First, they output probabilities
(in contrast to “exact knowledge”). Second, configuring these
outputs involves active choices of thresholds. Third, these
choices entail trade-offs between different evaluation measures.
At the end of the demonstration session, the students were
divided into two groups and assigned the problem-centered task
of helping “WeldCorp” use ML as an instrument to assess the
quality of their welding joints. A rubric for the task provided a
backstory, stating that WeldCorp was launched in 1994 in
Gothenburg, and subsequent expansion to other Swedish cities
led to the CEO experiencing problems with maintaining quality
control. So, s/he is now turning to ML for this purpose. The
rubric then told the students:
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
61
Your assignment is to help WeldCorp sustain its growth by
leveraging machine learning. Specifically, your task is to
analyze welding images (images of good and bad welding
points) to develop a model – using the no-code AI platform –
that can be useful for WeldCorp in a quality assurance context.
1. Describe and justify your choices regarding the data
processing, problem selection, and model training in
the no-code AI platform.
2. Describe how you evaluated your model’s predictions.
Are they accurate enough to use live for WeldCorp?
Why/why not?
3. Discuss: What could be done by WeldCorp to improve
the model’s results? How would they implement this
type of solution in their business?
An important aim during this assignment was to prompt
students to think about and justify their choices during training,
and the output of their model(s), rather than simply striving to
optimize the performance of the model(s). As the module is a
part of an AI business course, we also wanted the students to
discuss how WeldCorp could integrate AI into their
organization.
The students were divided into two groups. The start of the
course included a presentation exercise in which the students
were asked to state their names and educational background. As
two of the students had experience in computer science, we
intentionally placed these students in separate groups. To get
the students started, they were given a small dataset of 157
images of good and bad welds. The groups were then given
enterprise accounts providing access to the no-code AI
platform. Before engaging in a similar project, we advise
instructors to carefully assess the kinds of user configurations
that candidate platforms offer, as their user management
options vary, and potential issues must be addressed before the
students attempt to use them.
Five days after the initial workshop, a Q&A session was
held with the student groups. No instructions were given before
this session and the content was largely based on the students’
queries. Most questions concerned data. This was consistent
with expectations, as models trained using the intentionally
limited dataset handed out during the previous session would
perform badly, regardless of the platform settings that the
students chose. As already mentioned, data collection and
processing play a key role in ML, and “there is no AI without
data” (Gröger, 2021, p. 98). Illustrative queries from the
students concerned the quality of the supplied dataset, tentative
workarounds, and image formats. However, the main
conclusion the students drew was that more data was needed to
train a model that would produce relevant results.
Between the Q&A and final seminar, the students were
supposed to email or book appointments with the responsible
teachers if they needed supervision. The teachers could observe
and aid the students as they uploaded data and then trained and
evaluated ML models. After the Q&A session, we observed
how the students engaged in data collection and uploaded larger
datasets with various images to the platform. As the students
aimed to train models based on a binary classification of good
and bad welds, they needed two labels (“good” and “bad”). The
students applied the procedures previously demonstrated to
them, trained several models, and iteratively fine-tuned the
platform settings, using several sources of data, including social
media, Google image search, and Kaggle.
While the workshop and Q&A session were held on
campus, the final seminar was held via Zoom (January 10,
2022) as this was during a time when staff and students at higher
education institutions were gradually returning to campus after
the COVID-19 pandemic. The written assignment included the
following instructions:
You will be presenting your results both in the form of a
short paper, max ten pages, and orally in the final seminar.
During the seminar, each group will get 30 minutes to present
their results. You must also participate actively by answering
questions and comments regarding the presentation. Your short
paper should begin with a cover page on which you state the
names of the group participants, the name of the course, and
the semester. It is to be handed in at the start of the seminar.
During discussions in a final seminar, the students were
encouraged to reflect upon the ML process, to enable them to
integrate their acquired skills. In addition to discussing the ML
workflow, the students also proposed ideas for operationalizing
their work in a live setting, such as using automated cameras to
feed data on welding points for evaluation by the DL model. In
this seminar, the teachers mainly played a facilitating role, as
the students posed questions and reflected on their results. The
students received pass or fail grades for the task. To pass they
needed to:
• Present a logically coherent suggestion for WeldCorp,
both in writing and orally during the seminar.
• Formulate results and associated discussion in a
grammatically correct way and with consistent use of
concepts and terms.
The teaching activities outlined above are linked to the five
instruction principles and summarized in Table 3. Depending
on the course, and available data and case(s), these activities
can be varied. For example, the workshop can be divided into
two separate events, with an initial lecture focusing on
theoretical aspects of ML, followed by a more hands-on
workshop. Moreover, the group case can be presented as an
individual or pair-wise task, although this might neglect the
collective character of data work.
5.2 The Benefits of Using No-Code AI in Education
This subsection presents the observed benefits of using no-code
AI to teach ML, which are described below and summarized in
Table 4.
5.2.1 Benefit 1: Visualization of Data and Provision of a
Graphical Interface for Uploading Data. As already
mentioned, a crucial and time-consuming part of working with
ML is collecting and processing data. As the no-code AI
platform automated many parts of the ML workflow, students
had time to spend during the exercise on consideration and
labeling of the data. This was an anticipated and important part
of the task, especially as previous studies have highlighted
tensions among people involved in labeling data for supervised
learning (Lebovitz et al., 2021).
In their course evaluations and written feedback, the
students heavily emphasized an increase in their awareness of
the importance of data, and how the no-code approach enabled
them to focus on important features of the datasets used,
potential flaws in them, and problem-solving rather than model-
optimization, as illustrated by the following three quotations:
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
62
“I’ve obtained practical knowledge and experience of the
impact of data. And I’ve seen the impact of flaws in the dataset
first-hand. Thus, I think this was an optimal learning method
considering our (and my) educational background.” – Student
Evaluation.
“[I’ve learnt] that data matters! The choice, generating and
cleansing of data is crucial.” – Student Evaluation.
“For me, the barrier to understanding the practical use of AI
(or to ever try it myself) has been my lack of programming and
coding skills. With the no-code approach, I got the opportunity
to try experiments and thus got a “black-boxed” grasp of how it
works. With that, I could focus on the problem that I wanted to
solve, the learning dataset and its effect on the results, and also
on the result itself. So, I think I learned more about AI in this
course than I have in all the other courses combined, and that is
without any code.” – Student Evaluation.
Both groups chose to label their images in a binary fashion
as “good” or “bad.” To establish the consensus required for
creating “ground truths,” one of the groups formalized the data
labeling process in their report with a “weld quality
framework.” The other group strongly engaged in data
augmentation as they extended their dataset 4 to 5-fold by
manipulating the images by zooming, cutting, and rotating
them. These slightly different approaches were displayed in the
results and reflected upon in the student reports. While the
group that applied data augmentation focused more on the
performance of the models they created, and thus achieved
better measures (lower rates of false positives or negatives), the
other group focused more on trying to explain the output of the
models they created, i.e., why the models made certain
predictions.
5.2.2 Benefit 2: Access to a Portfolio of Pre-Trained Models,
Tutorials, and Datasets, as Well as the Automatic Selection
and Fine-Tuning of Algorithm(s) for Training. Both groups
ended up using a pre-trained model (EfficientNetB0) to solve
an image classification problem (single label) in the platform.
Each group formed training, validation, and test sets,
respectively, containing 80%, 10%, and 10% of their full
datasets (images), which is common practice and a default
option in the platform. The students refined their models’
outputs in two ways. First, they iteratively adjusted settings in
the platform, such as increasing the training rate (with careful
monitoring of the variances of performance measures of the
predictions generated by splitting the dataset to avoid
overtraining the model). The platform assists such adjustment
by suggesting settings to enhance the models’ performance,
e.g., switching to a different pre-trained model, and modifying
the learning rate (Figure 4).
Figure 4. Suggestions to Improve Model
Performance
Principle
Description
Problem-
centered
learning
The students were presented with a
case of a welding company,
WeldCorp, seeking to expand and
scale up its business while improving
quality control. To help these efforts
they were encouraged to apply ML to
differentiate between good and bad
weld points.
Activation
Since the students had diverse
educational backgrounds (business
and administration, computer science,
and behavioral science), we chose to
use a no-code AI platform. This
enabled them to incorporate previous
skills and work during the course,
even if they lacked previous
experience in data science.
Demonstration
We showed the students several
examples of ways to train ML models
via the no-code AI platform. Students
were encouraged to take tutorials and
experiment with different types of
open datasets (e.g., table, text, and
image-based), and problems that can
be accessed through the platform.
Application
The students were divided into two
groups, and each student was given
access to an enterprise account
enabling them to use the no-code AI
platform to address a new type of
problem by applying the previously
demonstrated procedures.
Integration
Students were encouraged to reflect on
their learning during the final seminar
in both a survey and the course
evaluation. During the final seminar,
they were also expected to learn from
each other by preparing questions for
the other group.
Table 3. Activities That We and the Students
Engaged in, Linked to the Five Principles of
Instruction
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
63
5.2.3 Benefit 3: Visual Interface for Evaluating and
Comparing the Performance of Models (e.g., Through ROC
Curve- and Confusion Matrix-Based Analyses). Second, as
particularly strongly emphasized by one of the groups, the
students strove to ensure the data included were contextually
relevant and suitable for WeldCorp’s purposes. This was done
after they received output from the ML model in the form of
confusion matrices and ROC curves (Figures 5 and 6) and could
assess whether certain types of images were incorrectly
classified, identify potential biases in the data, and signs of
model overtraining. Examples mentioned during the final
seminar were images of painted welds, which would not be
relevant in the industrial context they imagined.
Figure 5. Illustrative Model Evaluation Output
Figure 6. ROC-Curve From One of the Student
Reports
Available features briefly mentioned in the course included
tools to deploy the models created in the platform. This was not
relevant to the assigned task, as the students were not expected
to integrate their solution in a live environment; we presented a
few paths to do so. Examples included plugins for common
software (such as Excel, Google Sheets, and Bubble) and the
ability to call APIs for easy integration of a model in an
operating environment. The platform also includes a graphical
interface for predicting new images, as shown in Figure 7. We
used this function during the final seminar to show the students
how their models performed on selected images of good and
bad welds.
Figure 7. Results of a Test of a Model’s Performance on
Unseen Data During the Final Seminar
Thus, by simplifying parts of the ML workflow related to
training, evaluating, and deploying models, learners can focus
on data collection and interpreting outputs of the models to gain
a sense of whether the chosen approach is suitable and feasible
rather than engaging in model optimization. Based on our
materials, we generated themes in the form of distinct ways that
no-code AI facilitates learning about ML. These themes are
described in Table 4.
ML workflow
Role of no-code AI
Data
collection /
preparation
Provision of a graphical interface for
visualization, uploading, and
processing data.
Model training
Access to a portfolio of pre-trained
models, tutorials and datasets, as well
as automatic selection and fine-tuning
of one or more algorithm(s) for
training.
Evaluation
Visual interface for evaluating and
comparing the performance of models
(e.g., through ROC curve- and
confusion matrix-based analyses).
Deployment
APIs with complementary plugins to
aid integration in organizational
settings.
Table 4. How No-Code AI Can Facilitate Learning
About ML
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
64
5.3 Challenges With Using No-Code AI in Education
Our approach was not free of challenges, including three
summarized here. First, it is important to formulate a live case
in terms of ML and make a preliminary judgement of the
feasibility of the students collecting the necessary data during
the task. Finding an appropriate case may be time consuming,
but data repositories, such as Kaggle, may aid this process.
Second, as mentioned, the teachers also encountered challenges
related to user management routines before the module started
and needed help from the platform owners to set up separate
organizations for the students. These challenges highlight the
importance of considering and addressing potential user
management issues in advance and choosing an appropriate
platform for the intended purposes. The market for these
platforms is rapidly evolving. While the Peltarion platform is
now discontinued, several alternatives are available, such as
BigML, HuggingFace, and solutions from large tech companies
(e.g., Microsoft Azure, Amazon SageMaker, Google AutoML,
and Teachable Machine). These often come in both free and
paid versions. For individual use, the free versions may be
suitable for smaller tasks and datasets. A common advantage of
paid versions is the incorporation of more collaborative
features, which enables re-use and comparisons of student
projects over the years. Whichever platform and version is
chosen, it is also important to ensure that students do not upload
sensitive data, depending on the regulatory context of the
educational setting. Third, the student feedback included
proposals that groups should be smaller in future versions of the
course, as they experienced difficulties in engaging everyone
simultaneously when using the platform.
6. CONCLUDING DISCUSSION
As the no-code approach enabled students to engage in
collective data work the selected empirical setting provided an
ideal opportunity to address our two questions:
RQ1: How can no-code AI be used to teach ML in non-
technical educational programs?
RQ2: What are the benefits and challenges of using no-
code AI in education?
We answer RQ1 by proposing a problem-centered approach
to using no-code AI in higher education, with instruction to
teachers. Regarding RQ2, we show how no-code AI can help to
guide students through the ML workflow (data processing,
model training, evaluation, and deployment), and present
important challenges (ML case construction, platform selection
and user management, and student group composition) that we
encountered during the course.
Our contribution to the IS education literature is two-fold.
First, we provide information for instructors on how to
incorporate no-code AI in their courses. Second, we provide
insights into the benefits and challenges of using no-code AI
tools to support the ML workflow in educational settings.
Through this study, we have set the stage for incorporating
a new generation of AI tools in IS curricula by showing how
they can be used to support students in analyzing live cases,
particularly in conjunction with an approach based on
principles of instruction. By doing so, in this paper, we have
proposed an innovative solution to an IS teaching need,
grounded in theory and tested in an educational setting
(Lending & Vician, 2012). The novelty of our approach is the
application of tools that are usually only accessible to computer
scientists to problems related to business practices and
phenomena addressed in social sciences. As the no-code AI
tools available are rapidly increasing and evolving (a few, of
many, examples of contemporary no-code or low-code
solutions that support the ML workflow include BigML,
Huggingface, and Teachable Machine), we urge educators to
keep track of this development and find approaches to
implement such tools in their curricula, in combination with
lessons on how to use AI in effective and responsible ways.
7. REFERENCES
Bhattacharyya, S. S., & Kumar, S. (2021). Study of
Deployment of “Low Code No Code” Applications Toward
Improving Digitization of Supply Chain Management.
Journal of Science and Technology Policy Management,
14(2), 271-287. https://doi.org/10.1108/JSTPM-06-2021-
0084
Bishop, C. M. (2006). Pattern Recognition and Machine
Learning. New York, NY: Springer.
Braun, V., & Clarke, V. (2012). Thematic Analysis, In H.
Cooper, P. M. Camic, D. L. Long, A. T. Panter, D.
Rindskopf, & K. J. Sher (Eds.), APA Handbook of Research
Methods in Psychology, Vol. 2. Research Designs:
Quantitative, Qualitative, Neuropsychological, and
Biological (pp. 57-71). American Psychological
Association.
Chapman, P., Clinton, J., Kerber, R., Khabaza, T., Reinartz, T.,
Shearer, C., & Wirth, R. (1999). The CRISP-DM User
Guide. 4th CRISP-DM SIG Workshop in Brussels in March
(Vol. 1999).
Chen, L. (2022). Current and Future Artificial Intelligence (AI)
Curriculum in Business School: A Text Mining Analysis.
Journal of Information Systems Education, 33(4), 416-426.
Coffin, E. (2021). I Think I Need AI! What Is AI? BNP Media.
Clarke, V., & Braun, V. (2014). Thematic Analysis. In
Encyclopedia of Critical Psychology (pp. 1947-1952).
Springer, New York, NY. https://doi.org/10.1007/978-1-
4614-5583-7_311
Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs,
C., Crick, T., Duan, Y., Dwivedi, R., Edwards, J., Eirug, A.,
Galanos, V., Ilavarasan, P. V., Janssen, M., Jones, P., Kar,
A. K., Kizgin, H., Kronemann, B., Lal, B., Lucini, B.,
Medaglia, R., Meunier-FitzHugh, K. L., Meunier-
FitzHugh, L. C. L., Misra, S., Mogaji, E., Sharma, S. K.,
Singh, J. B., Raghavan, V., Raman, R., Rana, N. P.,
Samothrakis, S., Spencer, J., Tamilmani, K., Tubadji, A.,
Walton, P., Williams, M. D. (2021). Artificial Intelligence
(AI): Multidisciplinary Perspectives on Emerging
Challenges, Opportunities, and Agenda for Research,
Practice and Policy. International Journal of Information
Management, 57.
https://doi.org/10.1016/j.ijinfomgt.2019.08.002
Fredriksson, T., Mattos, D. I., Bosch, J., & Olsson, H. H.
(2020). Data Labeling: An Empirical Investigation Into
Industrial Challenges and Mitigation Strategies.
International Conference on Product-Focused Software
Process Improvement (pp. 202-216). Springer, Cham.
https://doi.org/10.1007/978-3-030-64148-1_13
Geske, F., Hofmann, P., Lämmermann, L., Schlatt, V., &
Urbach, N. (2021). Gateways to Artificial Intelligence:
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
65
Developing a Taxonomy for AI Service Platforms.
European Conference on Information Systems (ECIS).
Gröger, C. (2021). There Is No AI Without Data.
Communications of the ACM, 64(11), 98-108.
https://doi.org/10.1145/3448247
Hilbert, M., & López, P. (2011). The World’s Technological
Capacity to Store, Communicate, and Compute
Information. Science, 332(6025), 60-65.
https://doi.org/10.1126/science.1200970
Holmström, J., Sandberg, J., & Mathiassen, L. (2011).
Educating Reflective Practitioners: The Design of an IT
Management Masters Program. Americas Conference on
Information Systems 2011 Proceedings. 247.
https://aisel.aisnet.org/amcis2011_submissions/247
How, M. L., Chan, Y. J., Cheah, S. M., Khor, A. C., & Say, E.
M. P. (2021). Artificial Intelligence for Social Good in
Responsible Global Citizenship Education: An Inclusive
Democratized Low-Code Approach. Proceedings of the 3rd
World Conference on Teaching and Education (pp. 81-89).
https://doi.org/10.33422/2nd.worldcte.2021.01.08
Humble, N., & Mozelius, P. (2022). The Threat, Hype, and
Promise of Artificial Intelligence in Education. Discover
Artificial Intelligence, 2(1), 1-13.
https://doi.org/10.1007/s44163-022-00039-z
Jordan, M. I., & Mitchell, T. M. (2015). Machine Learning:
Trends, Perspectives, and Prospects. Science, 349(6245),
255-260. https://doi.org/10.1126/science.aaa8415
Kaggle (2023). https://www.kaggle.com/
Kayhan, V. (2022). When to Use Machine Learning: A Course
Assign ment. Communications of the Association for
Information Systems, 50, 122-142.
https://doi.org/10.17705/1CAIS.05005
Kelleher, J. D., & Brendan, T. (2018). Machine Learning 101.
In Data Science (pp. 97-150), MIT Press.
https://doi.org/10.7551/mitpress/11140.003.0008
Kling, N., Runte, C., Kabiraj, S., & Schumann, C. A. (2022).
Harnessing Sustainable Development in Image Recognition
Through No-Code AI Applications: A Comparative
Analysis. International Conference on Recent Trends in
Image Processing and Pattern Recognition, 146-155.
Springer, Cham. https://doi.org/10.1007/978-3-031-07005-
1_14
Kühl, N., Schemmer, M., Goutier, M., & Satzger, G. (2022).
Artificial Intelligence and Machine Learning. Electronic
Markets, 32, 2235-2244. https://doi.org/10.1007/s12525-
022-00598-0
Leavitt, K., Schabram, K., Hariharan, P., & Barnes, C. M.
(2021). Ghost in the Machine: On Organizational Theory in
the Age of Machine Learning. Academy of Management
Review, 46(4), 750-777.
https://doi.org/10.5465/amr.2019.0247
Lebovitz, S., Levina, N., & Lifshitz-Assaf, H. (2021). Is AI
Ground Truth Really True? The Dangers of Training and
Evaluating AI Tools Based on Experts’ Know-What. MIS
Quarterly, 45(3b), 1501-1525.
https://doi.org/10.25300/MISQ/2021/16564
LeCun, Y., Bengio, Y., & Hinton, G. (2015) Deep Learning.
Nature, 521(7553), 436-444.
https://doi.org/10.1038/nature14539
Lending, D., & Vician, C. (2012). Writing IS Teaching Tips:
Guidelines for JISE Submission. Journal of Information
Systems Education, 23(1), 11-18.
Leidner, D. E., & Jarvenpaa, S. L. (1995). The Use of
Information Technology to Enhance Management School
Education: A Theoretical View. MIS Quarterly, 19(3), 265-
291. https://doi.org/10.2307/249596
Lethbridge, T. C. (2021, October). Low-Code Is Often High-
Code, So We Must Design Low-Code Platforms to Enable
Proper Software Engineering. International Symposium on
Leveraging Applications of Formal Methods (pp. 202-212).
Springer, Cham. https://doi.org/10.1007/978-3-030-89159-
6_14
Liebowitz, J. (1992). Invited Paper: Teaching an Applied
Expert Systems Course: A Content Outline. Journal of
Information Systems Education, 4(3), 5-10.
Liebowitz, J. (1995). Integrating Expert Systems Throughout
the Undergraduate Curriculum. Journal of Information
Systems Education, 7(1), 34-36.
Lins, S., Pandl, K. D., Teigeler, H., Thiebes, S., Bayer, C., &
Sunyaev, A. (2021). Artificial Intelligence as a Service.
Business & Information Systems Engineering, 63(4), 441-
456. https://doi.org/10.1007/s12599-021-00708-w
Luan, H., & Tsai, C. C. (2021). A Review of Using Machine
Learning Approaches for Precision Education. Educational
Technology & Society, 24(1), 250-266.
Luo, Y., Liang, P., Wang, C., Shahin, M., & Zhan, J. (2021,
October). Characteristics and Challenges of Low-Code
Development: The Practitioners’ Perspective. Proceedings
of the 15th ACM/IEEE International Symposium on
Empirical Software Engineering and Measurement (pp. 1-
11). https://doi.org/10.1145/3475716.3475782
Ma, Y., & Siau, K. (2019). Higher Education in the AI Age.
Americas Conference on Information Systems 2019
Proceedings. 4.
https://aisel.aisnet.org/amcis2019/treo/treos/4
Mathiassen, L., & Purao, S. (2002). Educating Reflective
Systems Developers. Information Systems Journal, 12(2),
81-102. https://doi.org/10.1046/j.1365-2575.2002.00122.x
Merrill, M. D. (2007). A Task-Centered Instructional Strategy.
Journal of Research on Technology in Education, 40(1), 5-
22. https://doi.org/10.1080/15391523.2007.10782493
Merrill, M. D. (2013). First Principles of Instruction:
Identifying and Designing Effective, Efficient and Engaging
Instruction. Hoboken, NJ: Pfeiffer/John Wiley & Sons.
Peltarion (2022). The Peltarion Deep Learning Platform.
Richardson, M. L., & Ojeda, P. I. (2022). A “Bumper-Car”
Curriculum for Teaching Deep Learning to Radiology
Residents. Academic Radiology, 29(5), 763-770.
https://doi.org/10.1016/j.acra.2021.11.016
Rokis, K., & Kirikova, M. (2022). Challenges of Low-
Code/No-Code Software Development: A Literature
Review. Proceedings of the International Conference on
Business Informatics Research (pp. 3-17). Springer, Cham.
https://doi.org/10.1007/978-3-031-16947-2_1
Russell, S., & Norvig, P. (2020). Artificial Intelligence: A
Modern Approach (4th Edition). Pearson. ISBN 978-
0134610993.
Sahay, A., Indamutsa, A., Di Ruscio, D., & Pierantonio, A.
(2020, August). Supporting the Understanding and
Comparison of Low-code Development Platforms. 46th
Euromicro Conference on Software Engineering and
Advanced Applications (SEAA) (pp. 171-178). IEEE.
https://doi.org/10.1109/SEAA51224.2020.00036
Journal of Information Systems Education, 35(1), 56-66, Winter 2024
https://doi.org/10.62273/CYPL2902
66
Sambasivan, N., Kapania, S., Highfill, H., Akrong, D., Paritosh,
P., & Aroyo, L. M. (2021, May). Everyone Wants to do the
Model Work, not the Data Work: Data Cascades in High-
Stakes AI. Proceedings of the 2021 CHI Conference on
Human Factors in Computing Systems (pp. 1-15).
https://doi.org/10.1145/3411764.3445518
Schröer, C., Kruse, F., & Gómez, J. M. (2021). A Systematic
Literature Review on Applying CRISP-DM Process Model.
Procedia Computer Science, 181, 526-534.
https://doi.org/10.1016/j.procs.2021.01.199
Simsek, Z., Vaara, E., Paruchuri, S., Nadkarni, S., & Shaw, J.
D. (2019). New Ways of Seeing Big Data. Academy of
Management Journal, 62(4), 971-978.
https://doi.org/10.5465/amj.2019.4004
Sturm, T., Gerlach, J. P., Pumplun, L., Mesbah, N., Peters, F.,
Tauchert, C., Nan, N., & Buxmann, P. (2021). Coordinating
Human and Machine Learning for Effective Organizational
Learning. MIS Quarterly, 45(3), 1581-1602.
https://doi.org/10.25300/MISQ/2021/16543
Sundberg, L. & Holmström, J. (2022). Towards ‘Lightweight’
Artificial Intelligence: A Typology of AI Service
Platforms. Americas Conference on Information Systems
2022 Proceedings. 13.
https://aisel.aisnet.org/amcis2022/sig_odis/sig_odis/13
Sundberg, L., & Holmström, J. (2023). Democratizing
Artificial Intelligence: How No-Code AI Can Leverage
Machine Learning Operations. Business Horizons.
https://doi.org/10.1016/j.bushor.2023.04.003
Umeå University (2022). AI for Business Course Curriculum,
https://www.umu.se/en/education/syllabus/2in408/
Yan, Z. (2021). The Impacts of Low/No-Code Development on
Digital Transformation and Software Development. arXiv
preprint arXiv:2112.14073.
Wang, S., & Wang, H. (2021). A Teaching Module of No-Code
Business App Development. Journal of Information
Systems Education, 32(1), 1-8.
Woo, M. (2020). The Rise of No/Low Code Software
Development—No Experience Needed? Engineering, 6(9),
960-961. https://doi.org/10.1016/j.eng.2020.07.007
AUTHOR BIOGRAPHIES
Leif Sundberg is an associate professor at the Department of
Informatics, Umeå University.
Sundberg’s research interests
involve digital government, the use
of no-code artificial intelligence, and
risk society studies. Sundberg has a
broad teaching experience in
engineering management and
information systems. He has
published his work in journals such
as Safety Science and Information Polity and presented it at
international conferences like IFIP EGOV-CeDEM-EPART
and AMCIS.
Jonny Holmström is a professor of information systems at
Umeå University and director and
co-founder of Swedish Center for
Digital Innovation. His research
interests are digital innovation,
digital transformation and digital
entrepreneurship. He is serving on
the editorial boards of CAIS, EJIS,
Information and Organization, and
JAIS. His work has appeared in
journals such as Communications of the AIS, Design Issues,
European Journal of Information Systems, Information and
Organization, Information Systems Journal, Information
Technology and People, Journal of the AIS, Journal of
Information Technology, Journal of Strategic Information
Systems, MIS Quarterly, Research Policy, and The Information
Society.
STATEMENT OF PEER REVIEW INTEGRITY
All papers published in the Journal of Information Systems Education have undergone rigorous peer review. This includes an
initial editor screening and double-blind refereeing by three or more expert referees.
Copyright ©2024 by the Information Systems & Computing Academic Professionals, Inc. (ISCAP). Permission to make digital
or hard copies of all or part of this journal for personal or classroom use is granted without fee provided that copies are not made
or distributed for profit or commercial use. All copies must bear this notice and full citation. Permission from the Editor is
required to post to servers, redistribute to lists, or utilize in a for-profit or commercial use. Permission requests should be sent to
the Editor-in-Chief, Journal of Information Systems Education, editor@jise.org.
ISSN: 2574-3872 (Online) 1055-3096 (Print)
