Integrating Generative AI to

Augment Student Thinking and

Collaboration
Bodong Chen
Associate Professor, Penn GSE
About me

● Learning scientist and educational technologist by training

● Directs Penn GSE Wonder Lab and Knowledge Building
Innovation Network
● Tool designer and developer (Knowledge Forum, DataX,
learning analytics dashboards…)
● World traveler – Southwestern China, Beijing, Toronto, Twin
Cities, Philadelphia…
● Educator at heart
Overview of the session

● The evolving AI landscape

● Intelligence augmentation
● Group discussion
● Some provocative propositions
● A case study
● Work ahead
Key messages:

● AI has been around for a while.

● Shift from AI to ‘IA’.
● Systems rather than disruptions.
● Educators are central!
How capable is AI?
Open https://chat.openai.com/

Play
● “Tell me a joke about [...]”
● “Write an essay about [...]”
● “How to write a Python script to
[...]”
● “Draft a lesson plan on Columbus
Day for 9th graders”

● “I threw my cup against the wall. It

broke into pieces. What broke into
pieces?” Then replace wall with
mirror.
● “Where are the wildfires
happening in Canada?”
What is
ChatGPT?
(according to
ChatGPT
September 25
Version)
 Training Data of GPT-3

Brown, T. B., et al. (2020). Language Models are Few-Shot Learners (arXiv:2005.14165). arXiv.
http://arxiv.org/abs/2005.14165
Training: Generate training examples
Training: Machine learning
From GPT to ChatGPT

Reinforcement
Learning from Human
Feedback (RLHF)
From GPT to ChatGPT

Chat Interface

Reinforcement
Learning from Human
Feedback (RLHF)
AI History and Competing Paradigms
Symbolic Subsymbolic, connectionist
● Rule-based ● Stochastic

Mitchell, M. (2019). Artificial intelligence: A guide for Image Credit

thinking humans. Farrar, Straus and Giroux.
A 1960s chatbot therapist Chatbot based on GPT-3
The state-of-the-art
of AI in 2021

Wooldridge, M. (2021). A Brief History of Artificial

Intelligence: What It Is, Where We Are, and
Where We Are Going. Flatiron Books.
 A Wave of
Generative AI
(GenAI)
as of late Sept
Generative Fill in Photoshop
Comparing Physician and ChatGPT Responses on Quality and Empathy
(Ayers et al., 2023)
What does all of this
mean for learning and
education?
Shift the focus from AI to
IA – intelligence augmentation
Human intelligence
has always been
'augmented'.

(Engelbart, 1962)
 lo gy
e
u ag act odo ing
ang rtif eth rain
L A M T

H → H-LAM/T

To improve human capabilities, we improve

the system in which a human operates.
(Engelbart, 1962)
What is this?
 lo gy
e
u ag act odo ing
ang rtif eth rain
L A M T

H → H-LAM/T
Instead of asking how powerful ChatGPT is, ask:

How is the H-LAM/T system changed? And how

should it be changed?
Which parts of human experience are
augmented? How is the education
system augmented as a whole?

How is the augmentation guided by

theories of learning and teaching?
Group Discussion (15 min)

● What capabilities of ChatGPT (or other AI tools) have you

recognized or played with?
● What opportunities does ChatGPT offer for your context?
● What challenges or risks of using ChatGPT do you
anticipate?
Take notes: https://padlet.com/cbd_upenn/cpl_genai_oct03
Augmenting Student Thinking and
Collaboration with GenAI
Creative Ideas
https://zenodo.org/record/8072950
 The alignment
problem

● Side-effects (besides benefits) for

learning and learners
● Challenges to the teaching profession

OpenAI. (2023, March). GPT-4 System Card. OpenAI.

https://cdn.openai.com/papers/gpt-4-system-card.pdf
Proposition 1: To augment learning with GenAI
requires redesigning the system and adjusting
relations among parts of the system.
 Mindful engagement
with intelligent
technologies

Salomon, G., Perkins, D. N., & Globerson, T. (1991). Partners in

Cognition: Extending Human Intelligence with Intelligent
Technologies. Educational Researcher, 20(3), 2–9.
Proposition 1: To augment learning with GenAI
requires redesigning the system and adjusting
relations among parts of the system.
● How does mindful engagement with GenAI look like?
● What level of AI literacy is needed by students and teachers?
● How should classroom practices adjust to the ‘participation’ of GenAI?
● …
Effects with vs. of GenAI Systemic factors

Created with DALL-E 2 (Prompt: student and (Zhao & Frank, 2003)
artificial intelligence working together)
Proposition 2: GenAI can better augment learning if
we precisely pinpoint the intelligence that is being
augmented, and how the system is transformed.
● What learning processes are appropriate for augmentation by GenAI?
● How is agency distributed between human and GenAI?
● How should other parts of the system (such as definitions of learning)
adjust?
● …
Autodesk Dreamcatcher

Video, Mixed-Initiative Creative Interfaces

CoAuthor (Lee et al., 2022; https://coauthor.stanford.edu/)

(Shibani et al., 2023)

Proposition 3: The development of AI literacy is a
necessary condition for mindful engagement with
GenAI.
● What aspects of AI literacy are crucial for learners and teachers?
● How should AI literacy be taught in my school? Where to start?
● How to support the development of AI literacy among educators?
● …
What is AI literacy? (Long & Magerko, 2020)

Source
 Work with AI black boxes

Bearman, M., & Ajjawi, R. (2023). Learning to work with the black
box: Pedagogy for a world with artificial intelligence. British
Journal of Educational Technology.
https://doi.org/10.1111/bjet.13337
Principles of “effective ethical engagement” with GenAI

(Buckingham Shum, 2023)

 A Case Study
Bodong Chen, Xinran Zhu, Fernando Díaz del Castillo
Goals:
● Augmenting student creative work with ChatGPT (Proposition #2)
● Facilitating students’ AI literacy (Proposition #3)
● Seeking to redesign knowledge building environments (Proposition #1)

Context:
● A high-school class about World Religions course
● One teacher and 10 high school students
● Students were expected to examine various religions in the world and
develop religious literacy
 Design Two phases of using ChatGPT in the class:
- Exploring the problem space (Pattern 1)
● Co-design partnership between - Co-creating knowledge to be presented in the
the teacher and researchers final essay (Pattern 2)

● Situate ChatGPT in Knowledge-

Creating Dialogue Moves:

○ Problem definition
○ New ideas
○ Promisingness evaluation
○ Meta-dialogue
○ Comparison
○ Critical discourse
○ Higher-level ideas
(Bereiter & Scardamalia, 2017)

Note. Numbers indicate the steps within each design pattern.

Example ChatGPT Prompts Designed by Mr. F.
Essay Generation Prompt (Mr. F. gave ChatGPT 3.5 the same prompt he
gave the students, without any tailoring for the A.I.)
Write a brief essay that is no more than 600 words long including titles and
inline references, but not the bibliography. Your essay should address the
following issues based on the readings and discussions we have had in class
so far, and it must include complete in-line and bibliographic references to
the authoritative sources. You must include the readings addressed in class
along with any other sources you use. The sources are:
…
Key questions and issues to address in your essay:
What is religion (and what is not)?
What are the big ideas of religion (those topics or key areas that many
religions have in common and we would to address to understand them)?
What are some key questions about these big ideas that you are most
interested in?
As with any paper, yours should include a title, introduction, body,
conclusion and bibliography.
Steering Prompt
You are a teaching assistant in a high school-level introduction to world
religions course.
Students have read chapters in the book "God Is Not One" by Stephen
Prothero about specific religions and they have done some independent
research online. They will ask you questions to advance their understanding
of class topics and their own questions. Answer following this protocol
strictly:
1. Provide a brief answer in accessible language for 16-18 year olds,
assuming knowledge of themes touched on by Prothero in his book.
2. Highlight disagreements or different points of view on the issue that
bring nuance to the discussion.
3. Follow-up with a question that may help the learners understand nuances
and complexities of the issue discussed.
Are you ready for a question?
Research Questions
1. In what ways did students incorporate ChatGPT in
their knowledge building?
2. To what extent was students’ AI literacy enhanced,
and how?

Data Sources
● Primary data: semi-structured
interviews with 10 students (in
groups of 2 or 4)
● Secondary data:
○ Student writing and
artifacts generated from KF
and Miro
○ Teaching planning docs
and reflective journals
Data Analysis
● Interview data: iterative coding processing
○ Coding scheme:
■ Use AI in Knowledge Building: utility,
process, challenge, coping strategy
■ AI Literacy: mechanism, strength,
weakness, risk, societal implications,
human-AI relationship
(Bearman & Ajjawi, 2023; Laupichler et al., 2023; Long
& Magerko, 2020; UNESCO, 2023)
● Leveraged secondary data to triangulate with the
interview results
Results: Knowledge Building with ChatGPT

Utility of ChatGPT
● Information search
● Accomplishing mundane learning tasks, such as grammar check

Processes of Using ChatGPT

● Integrating ChatGPT in writing and discussion processes
● Most students imposed limits, some fully relied on it
● Navigating challenges for prompt engineering
● “Fact-check” & proper citation
Results: Students’ AI Literacy

Mechanisms of ChatGPT

● All students recognized ChatGPT as an AI technology

● The majority had a rudimentary understanding
○ ChatGPT queries a database of sources in real-time
○ OpenAI developers uploaded files to the database
● A few students delved deeper: computer algorithms powering it &
safeguard mechanisms put in place to filter out harmful information.
Results: Students’ AI Literacy

Strengths:
● Interpreting user prompts
● Retrieving information efficiently
● Offering quick and clear responses
● Students perceived the information provided by ChatGPT to be rich and diverse,
representing different "facts'' that inspired them to generate new ideas
● Students appreciated how ChatGPT's responses were akin to those from a human

Weakness:
● Output quality: inaccurate or dated information, “black box”
● Limited cognitive capabilities
Results: Students’ AI Literacy
Risk and Societal Implications:
● In school: Potential abuse of AI in school
settings, especially for students who might
not have the opportunities to learn and
understand how to use AI properly.
Relationship with ChatGPT:
● A valuable tool that greatly supported their
learning processes.
● Primary source for information.

● Beyond schoolwork: Combined impact ● Not overly depend on it and preferred to

of AI and social media on their generation, set a limit on their usage.
such as cyber violence and misinformation
● Most students did not fully trust AI.
● Bias: ChatGPT is not biased because of
the AI's strict role in responding solely to
prompts without any feelings or opinions.
Takeaways from the case study

● Student understanding and AI literacy developed in tandem

● The use of ChatGPT made learning harder, instead of
easier, as mindful engagement was needed
● Students took high-level responsibility in the long process
● The teacher was a co-learner of ChatGPT and played an
instrumental role in guiding students’ use of ChatGPT
Looking ahead

How do we integrate How do we engage As society becomes

GenAI in learning students and educators in ‘AI-mediated’, how are
scenarios? co-designing this radical learning and various
shift? relationships involved in
learning redefined?
What do we mean by How do designs and
GenAI literacy and how outcomes translate across
do we scaffold it? contexts?
Educators are central
to the solutions.
Thanks!
Bodong Chen

cbd@upenn.edu
bodong.ch