This is the slides from a webinar I gave to the senate of Universiti Padjajaran, Inodonesia as part of the activities in discussing on AI implications in education at their institution.
1. Artificial
Intelligence for
Teaching and
Learning
Assoc. Prof. Dr. Nurfadhlina Mohd Sharef
Deputy Director
Centre for Academic Development and Leadership Excellence (CADe-Lead)
Universiti Putra Malaysia
nurfadhlina@upm.edu.my
2. Assoc. Prof. Ts. Dr.
Nurfadhlina Mohd Sharef
Intelligent Computing Research Group, Faculty of Computer Science and Information
Technology, Universiti Putra Malaysia.
nurfadhlina@upm.edu.my, 0126672504, https://sites.google.com/view/nurfadhlina
Affiliation:
1. Associate Professor, Department of Computer Science,
Faculty of Computer Science and Information Technology,
Universiti Putra Malaysia.
2. Deputy Director (Innovation in Teaching and Learning),
Centre for Academic Development (CADe), Universiti Putra
Malaysia.
3. Task Force Member, National Artificial Intelligence
Roadmap Implementation 2021-2025, MOSTI
4. Member, Malaysia Council for e-Learning Heads at Public
University
5. Member, New Horizon in Science and Technology A New
Horizon for STI – A Strategy to Enhance Higher Education
in Malaysia (NHSTI) Expert Group, Ministry of Higher
Education Malaysia
6. Task Force Member, Health Workforce Culture Survey
Analytics, MOSTI
7. Interim Research Associate, Malaysia Institute for Ageing
Research (MyAGEING), Universiti Putra Malaysia.
8. Research Associate, Institute for Mathematical Research
(INSPEM), Universiti Putra Malaysia.
9. Secretary, Young Scientist Network-Academy of Sciences
Malaysia (YSN-ASM)
10. Chair, COVID-19 ASM Data Scientist Group, Academy of
Sciences Malaysia, MOSTI
Data Analytics
1. Digitalisation and IoT for Precision
Biodiversity
2.COVID19 Vaccination Distribution Planning
and Tracking
3. BSH- LHDNM Analytics Dashboard for
Program Bantuan Kerajaan
4. National Integrated Cybersecurity Threat
Factor Profiling
5. Learning Analytics and Chatbot for
Personalized Learning
Machine Learning
1. Deep Recurrent Q-Network Approach for
Multi Objective Recommendation System
2. Interactive Machine Learning based on Deep
Reinforcement Learning and Generative
Adversarial Network Hybrid for Digital Twin
Research Interests
● Artificial Intelligence
● Data Science and Data Analytics
● Text Mining and Question Answering
● Recommender Systems
● eLearning
Text Mining
1.Online Reputation Meter
2.Evolving Multi-Granular Temporal
Abstraction Method to Improve Clinical
Data Analysis
3.Multi-Tasking based Deep Learning for
Tweets Analytics
4.Deep Attention Model For Review-
based Multi-Criteria Recommendation
System
5.Sequence-to-Sequence Based Natural
Answer Generation Models
7. The allure of AI-powered tools to help
individuals maximize their
understanding of academic subjects
(or more effectively prepare for exams)
by offering them the right content, in
the right way, at the right time for them
✓ Enrich information discovery
experience
✓ Summarise and explain
✓ Answer questions
✓ Writing, composing and editing
✓ Increase productivity
We now live in GPT era…
Content
Type
Tool Application
Text Bard, ChatGPT,
ChatSonic, Claude,
Jasper AI
Conversation in question
answering
Image DALL-E, picsart, canva,
pixlr
Image generation
including photos and
artworks
Video Wibbitz, pictory,
Synthesia
Create short form
video online in minutes
Speech Whisper, Replicastudios AI voice actors
for games, film
& the metaverse
Audio Ampermusic, Veed, Murf Create your own songs
and compositions
7
8. can generate new text
based on the input they
receive
“GENERATIVE”
because they are
trained on a large
corpus of text data
before being fine-tuned
for specific tasks
"PRETRAINED"
because they use a
transformer based neural
network architecture to
process input text and
generate output text.
"TRANSFORMERS"
A Generative Pretrained
Transformer (GPT) is a
type of large language
model (LLM) that uses
deep learning to
generate human-like
text.
ChatGPT3 was
announced on
30/Nov/2022.
The magic
of ChatGPT
Centre for Academic Development (CADe), UPM
9. The use of ChatGPT in teaching and learning
1. Create lesson plans
2. Question answering
3. Text classification
4. Get fresh creative ideas or advise for a refined
thought
5. Create rubrics of assessments
6. Translate sentences
7. Compose a write up (e.g social media posts, product
review, promotional copywriting)
8. Summarize text
9. Text completion
10. Get generated text in a particular style (eg a 7-year
old understanding vs 27 years old)
9
You can keep on interacting
with it, ask it to refine and
personalise your request!
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
10. Centre for Academic Development (CADe)
Only (a) is
correct, but
ChatGPT got
it wrong, most
probably
because the
logic is wrong.
Both answers
are correct,
and ChatGPT
got it correct
Clustering is an unsupervised learning. Look at the
answer generated. According to the rules of Truth
table, yes ^ no = no. But in ChatGPT the reasoning
needs some work. This is an example how we as a
human educator could tune our way of assessing
students. Rather than asking straight forward fact
(which is lower level of Bloom taxonomy), we could
test their analysis level eg C4
10
, UPM
14. Centre for Academic Development (CADe)
14
AI in education is not new!!
, UPM
15. Three typical responses:
Ban ChatGPT
“Business as
usual”
Embrace
ChatGPT
The response we pick
must consider immediate
(course level – micro
picture) and future needs
(university level – macro
picture).
17. Advantage to Educators
● Help to get info
● Reduce time searching
● Use for future research ideas
● To write content and present research ideas
● Make education more interactive
● Reduce knowledge gaps
Concerns by Educators
● Worry students use it for FYP
● Discourage critical thinking
● Plagiarism
● Shortcut to assignments
● Concern to language teaching
● Laziness of reading and not creative
● Displacement of manpower
● Too dependent
● Redundant answer from student
● Would result to decrease in writing quality
● False fact
● Lack of citation
● Misuse
*Based on response to mentimeter during webinar by CADe-Lead, UPM on 2nd Feb
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
17
18. Safe and responsible use of AI in education?
Human-in-the-loop machine learning?
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
Technology is a facilitator for learning
Educators lead a crucial role in designing engaging
learning experience
A synergetic collaboration between multiple
entities
(e.g., the learner, the instructor, information, and
technology)
in the system is essential to ensure the learner’s
augmented intelligence
Cheating? Honesty? Truthfulness? Recency? Privacy?
Misleading? Manipulation?
18
20. machine-learning techniques behind generative AI
have evolved over the past decade…
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
Content Type Tool Application Implementation
Paradigm 1: AI-
directed
Learner as
recipient
Behaviorism Earlier work on
Intelligent Tutoring
Systems; ChatGPT
Paradigm 2: AI-
supported
Learner as
collaborator
Cognitive, Social
constructivism
Dialogue-based
Tutoring Systems;
Exploratory Learning
Environments;
ChatGPT
Paradigm 3: AI-
empowered
Learner as
leader
Connectivism, Complex
adaptive system
Human-computer
cooperation;
Personalised/adaptive
learning; ChatGPT
20
21. Guides for educators to embrace ChatGPT in activities
1. Allow students to use ChatGPT and have a discussion on the rules of its usage.
2. Practice retrieval and other memorisation activities that specify certain time, topic or activities
conducted previously to ensure students take effort to understand and analyze any references they have
utilized.
3. Create more collaborative and discussion activities. When students discuss, they do so from their own
working and long-term memory. Sure, they can look up quick answers, but to carry on a conversation, most
of the work comes from their own thinking. After a discussion, students can recap the discussion and share
their reflections about it ... and that's much harder to do with a bot.
4. Emphasize experiential learning and engage students in personalized elaboration that relate to their
local surroundings and routines. Let students demonstrate what they have learnt. Asking students to bring
in ideas, evidence, perspectives, and data from contemporary or personal events or geographical contexts
will make it more difficult (although not impossible) for them to just ask an AI to write their assignment.
5. Conduct activity that requires students to use ChatGPT to answer questions related to a topic, and
experiment to identify questions that can’t be answered. This will let the students think critically and
dive deeper into the topic.
21
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
23. Guides for educators to embrace ChatGPT in
1. Review your assessment; avoid straightforward questions or simple facts. Ask current, complex,
open, real-world problem, and based on a contextual topic or issues discussed in your lesson.
2. Be empowered with AI. Incorporate ChatGPT to personalise or draft a unique case study for
each student or their group based on their interest and level to use in your authentic assessment.
3. Include a section in assessments to let students to critique for improvements (what they got, if it
fits, how to organize it, how to communicate it effectively, etc.) and reflect their synthesis of the
information gathering through their reading, internet searching, peer discussions and ChatGPT
responses they have used and ask them to give their opinions and justifications.
4. Try different assessment types that are more immune to AI and can allow students to develop
and demonstrate understanding such as figurative related, oral assessment and live
demonstration. Additionally, staging assessments, such as requiring students to submit drafts,
receive feedback, and improve their work, are less prone to risk from generative AI.
5. Use ChatGPT to draft quiz questions and possible answers with feedback setting. Use
ChatGPT to generate a draft rubric that you can then refine.
23
Centre for Academic Development (CADe), UPM
31. Balancing the risk (for cheating) versus
opportunities (for feedback)
Developing meaningful and
relevant assessments are
more important than investing
in student surveillance
techniques.
In fact, ChatGPT can
encourage learning through
making mistakes and receiving
feedback iteratively
(productive struggle).
Do we reward our students for
effort or outcome?
Are we indirectly incentivizing
cheating?
→ We should not deprive our
students from learning values
and skills they will need as adults.
Wan Mohd Aimran Wan Mohd Kamil, UKM
32. For students, the biggest warning should be
that ChatGPT's "facts" cannot be taken as
such, and students should question every
piece of text they get from an AI.
ChatGPT has been known to deliver
inaccurate information, so students should
now – more than ever – be aware of the need
to verify information they receive through
different sources.
We can assume that more and more AI tools
will be developed to help educators get a
sense of whether a piece of text is AI-
generated or not.
32
Suggestion to
students
Disruption in education: Have we learned our lesson?
2020. Online learning
2022. Generative AI
What’s next on the horizon?
33. Example. ChatGPT can provide feedback to students:
Can ChatGPT substitute educators?
Submits a
draft
Generates
feedback
Utilizes
feedback
Supplies
rubric
Submits student
work
Utilizes
feedback
33
Potentially, ChatGPT as
a teaching assistant.
Wan Mohd Aimran Wan Mohd Kamil, UKM
34. Vicious versus virtuous loops
Generates
response
Give questions
Prompts Submits
Grades
Give questions
Generates
response
Prompts
Feedback
Submits-feedback-revises
Grades
Grades may not reflect student mastery
because ChatGPT short circuits student effort.
Grades reflect student mastery since ChatGPT
engages students in productive struggle.
34
Wan Mohd Aimran Wan Mohd Kamil, UKM
35.
36. CourseInfo
ProjectInfo
CommunityInfo
Impact
Teaching
Assessment
Semester
4.0
score
Faculty
Satisfact
ion
score Above
GIPP Yearly
2012 -
2014
Research
Grants
2012 –
RM342,000
2013 –
RM300,000
45 Research
2015 - 2016 2017 2018
Research
Grants
2015 – 2016
RM300,000
50 Research
Research
Grants
RM 450,000
31
Research
Research
Grants
RM 500,000
17
Research
Blended Learning
Semest
er
Blended
Learning
Percentage
Number of
course
81%
courses
achieved BL
status
2019
Exit
Surve
y
Semest
er
Score
2014-
2018
Trend
analysi
s
6.0
score
low
satisfactio
n
iPiCTL Yearly
Silver
Institutions
Categor
y
Bronze
Gold
CPD Monthly
Program
Manageme
nt
Experts
CourseName
#LocalParticipants
#IntParticipants
#Certificates
PutraMOO
C, MC UPM
Semeste
r
Score
Example ad-hoc analytics:
1. Blended Learning
Substitute
2. CGP, CGPA F2F vs
online
3. InnoCreative Delivery
4. Online Assessment
Various data has been generated related to academic development
purposes such as questionnaire and records. Typically the data is analysed
through descriptive analytics. This is sufficient for surface knowledge but
further analytic levels such as diagnostic, predictive and prescriptive allows
deeper knowledge and proactive decision making.
Education Analytics
@Centre for Academic Development (CADe), UPM
Themes
Faculty
SULAM Semeste
r
Innovatio
n
Adhoc
AR/
VR
Gamificatio
n
Mobile Apps
36
Micro-course
Macro-course
Digital badges
Students
MC UPM Continuous SULAM
37. Types of Learning Analytics
“What has already
happened?”
Visualization of
learning patterns
“Who could be at
risk?”
Prediction of low
achievement or engagement
“What should we do?”
Recommended action for
further teaching and
learning
Descriptive
Predictive
Prescriptive
Learning analytics is the
measurement, collection, analysis, and
reporting of data about learners and
their contexts, for purposes of
understanding and optimizing learning
and the environments in which it occurs
“Why something
happen?”
Hypothesizing factors of
performance and
satisfaction
Diagnostics
37
39. Learning Analytics
Forum
engagement
Quiz Attempts
Course
Completion
Time Spent
Learning
Learning
design
Grade
Choice of
Materials
✓
✓
✓
✓ ✓
✓ ✓
1) What are the characteristic
of excellent students?
2) What are the factors
contributing to teaching
assessment score?
3) What does courses with
high achievers have in
common?
4) How does courses across
groups, and semesters
differ?
5) What is the risk of students
to get poor results?
6) What are the achievements
of students in a micro-
course?
7) Which digital badge have
been awarded the most?
8) What are the factors
contributing to the
popularity of a micro-
course?
39
40. Learning Analytics Plan @UPM
Learning analytics data lake
● Ingest and integrate data for
learning (SMP, PutraBLAST) into 1
data lake
● Improvise online pedagogy to
ensure valid and consistent data
Course assessment analytics
● Identify of course learning outcome
achievements and gaps
● Conduct intervention manually in
the next semester, ensuring
students achieve all outcomes by
graduation
● Maintain/improve teaching
approach for the next cohort
Competency analytics
● Identify learners competency
completion in modules, micro and
macro courses
● Promote content to learners to
complete competency set
● Improve learning design
● Automated reminders
to students
Intelligent intervention and
personalized
recommendation
● Personalize advice to
students based on
potential risks
● Personalized delivery
and learning materials
for students
● Differentiated
assessments a.t.
competency and goals
1
2
3
4
42. Enjoying Learner Engaging Instructor Proactive Institution
1. Set goals through
pre-lesson
assessment
2. Engaged by human
instructor and
edubot
3. Receive
personalised
feedback
4. Get guide to
optimize
performance
5. Get personalized
learning plan and
materials
PPLA: a precision
personalized
learning advisor
plugin extension
in LMS through
HuML-based
prescriptive
analytics and
conversational
chatbot
1. Diagnose learner’s
progress,
participation,
satisfaction and
achievement
2. Monitor course
participation and
completion
3. Communicate and
give feedback to
learner
4. Elevate level or
customize delivery
5. Improvise
instructional design
Intelligent Assistant
1. Diagnose program
and student
completion
2. Communicate with
learner
3. Improvise
curriculum,
promotion and
student affairs
strategy
4. Improvise program
learning path and
system policy
1. Monitor learner’s
progress,
participation,
satisfaction and
achievement
2. Monitor course
participation and
completion
3. Communicate to
guide and remind
learner
4. Recommend
personalized
learning plan and
materials
@CSIT UPM
44. @IEEE Circuit and Signal
System Society (CASS)
Student info
hidden
intentionally
Learners
Worldwide
Students
competencies
accomplishment
Learners
classification
Lesson and
microlessons
completion
Badges
awarded
Learners
preferences
47. Enrolment, participation, popularity,
engagement, achievement, completion
analytics
Student advisory for personalized
learning plan recommendations
Continuous quality improvements -
learning design and curriculum content
@CADe UPM
48. UPM Learning Data Lake
1. Collect
2. Ingest
3. Blend
4. Transform
5. Publish
6. Distribute
@CSIT UPM, CADe UPM and IDEC UPM
Digital
cockpit
Informed
educator
Guided
students
49. GOVERNANCE, INFRA & INFO STRUCTURES
LEARNING ANALYTICS REVOLUTION (UPSCALE, AMPLIFY) FEASIBILITY
Strength Opportunity
Weaknesses Threat
1. Ready info structure and infrastructure
2. Historic knowledge base is ready @SMP,
@PutraBLAST
3. Database integration @SMP, @PutraBLAST is
possible with schema mapping
4. High readiness of students, lecturers and
administrators for AI-based apps
5. Lecturers and students have good skills and
acceptance of online teaching pedagogy
6. Various case studies been conducted
7. Good collaboration across entities
1. Upskilling in data analytics skills for Educator
4.0
2. Implementation of data governance
3. Implementation of intelligent assistant for
students, lecturers and administrators
4. Open Edumetric Dashboard for supporting
decision making
5. Data must be evaluated according to its
interoperability, reusability, usefulness,
applicability, appropriateness and effectiveness
so that it can be used in analytics
1. Systems are isolated and requires some
modules improvements
2. Teaching pedagogy and delivery service
considers conventional approach by default
3. AI-based apps for teaching and learning
implementation is new to UPM
4. Skills and knowledge for learning analytics is
low-moderate
1. Enculturation of data-driven insights requires
buy-in, high integrity, access control and trust
2. Standardization of data-driven actions based on
predictive and prescriptive learning analytics
requires proper change management strategy
50. The world’s citizens need to understand what the impact of AI might
be, what AI can and cannot do, when AI is useful, when its use
should be questioned, and how it might be steered for the public
good (UNESCO International Forum on AI and the Futures of Education under the theme of
Developing Competencies for the AI Era).
This requires everyone to achieve some level of competency with regard
to AI, including knowledge, understanding, skills, and value orientation.
Together, these might be called ‘AI literacy’.
Source:
https://unesdoc.unesco.org/ark:/
48223/pf0000380602
2022
AI literacy comprises both data literacy, or the ability to
understand how AI collects, cleans, manipulates, and analyses
data; and algorithm literacy, or the ability to understand how AI
algorithms find patterns and connections in the data, which might
be used for human-machine interactions.
AI Literacy =
Data Literacy +
Algorithm
Literacy
50
52. 1. Steer AI-and-education policy development and practices towards protecting human rights
and equipping people with the values and skills needed for sustainable development and
effective human-machine collaboration in life, learning and work;
2. Ensure that AI is human-controlled and centred on serving people, and that it is deployed
to enhance capacities for students and teachers.
3. Design AI applications in an ethical, non-discriminatory, equitable, transparent and
auditable manner; and monitor and evaluate the impact of AI on people and society
throughout the value chains.
4. Foster the human values needed to develop and apply AI.
5. Analyse the potential tension between market rewards and human values, skills, and
social well-being in the context of AI technologies that increase productivity.
6. Define values that prioritize people and the environment over efficiency, and human
interaction over human-machine interaction.
7. Foster broad corporate and civic responsibility for addressing the critical societal issues
raised by AI technologies (such as fairness, transparency, accountability, human rights,
democratic values, bias, and privacy).
8. Ensure that people remain at the core of education as an implicit part of the technology
design; and protect against automating tasks without identifying and compensating for the
values of current practices.
52
Centre for Academic Development and Leadership Excellence (CADe-Lead), UPM
54. Guide for educators
As an educator, it is important to have a general understanding of artificial
intelligence and how it can be used in education. Here are a few key points
to consider:
1. AI can be used to personalize learning experiences for students, by
adapting to their strengths, weaknesses, and learning style.
2. AI can be used to assist with grading and providing feedback on
student assignments.
3. AI can be used to create interactive and immersive learning
environments.
4. It is important to be aware of the potential ethical implications of
using AI in education, such as issues of algorithm bias and user
privacy.
5. As AI continues to advance and become more widely used in
education, it is important for educators to stay up-to-date on
developments and best practices for incorporating AI into their
teaching.
54
Centre for Academic Development (CADe), UPM