Asian American Pacific Islander Month DDSD 2024.pptx
UNESCO-Teaching and Learning with GenAI.pdf
1. Teaching and learning with generative AI
Fengchun Miao
Chief, Unit for Technology and AI in Education
Future of Learning and InnovationDivision, Education Sector, UNESCO
2. Supporting 80+ countries’ digital education policies and 5 guidance
Guidelines
on OER
Policies
AI and Education:
Guidance for
Policy-makers
Guidelines for
ICT in Education
Policies
Steering digital futures of education (Digital open school projects)
▪ 80+ examples on AI & ED:
Compendium (2019); (2020)
▪ First global mapping of
government-endorsed AI
curricula (2022)
Worldwide grass-rooted case studies
▪ 24 winning projects of UNESCO Prize for
ICT in Education
▪ Case studies on distance during COVID-19:
Finland, Korea, Saudi Arabia
▪ 14 best practices in mobile learning
▪ OER: Policy, costs,
and transformation
–14 frontrunner
projects on OER
Setting global standards on digital education
Guidance for
generative AI in
education and research
Beijing Consensus on AI and
Education
Recommendation on
the Ethics of AI
Recommendation on Open
Educational Resources
UNESCO’s main achievements in digital education
Global convening (e.g., Digital Learning Week, Gateway)
AI competency
frameworks:
teachers & students
Education
and
Blockchain
3. Beijing
Consensus
Guidancefor
policy-makers
Seminars for
50+ countries
AI & Education
policy
Watching AI
trends
Recommendation
on the Ethics of AI
Survey on AI in
education
GenerativeAI
AI competencies
for school students
Mapping K-12 AI
curricula
Report on “K-
12 AI curricula”
To be launched
in June2024
AI competencies
for teachers
Consultations Expert meetings
To be launched
in June2024
Guidance for generative AI in education and research
2019 Conference
(Beijing Consensus)
2020 Forum
(AI competencies)
2021 Forum
(AI as a
common good)
2022 Forum
(AI for teachers)
UNESCO Work on AI and Education
2023
(100+ ministers and 10 000+ participantsfrom 120+ countries)
Ministerial
Meeting on
generative AI
Digital
Learning
Week 2023
2024
Digital
Learning
Week 2024
A publication on “AI and Futures of Learning”
5 regional
seminars on
AI competency
for 100+
countries
4. Main publications on AI and Education
6 UN languages 6 UN languages + Korean 6 UN languages +
Portuguese
Will be in 10+ languages
Top 10 for 2 years
Top 20 for 2 years Top 30 for 2 years No.1 for 4 months
Ranking in downloads and consultationsin all UNESCOpublications
Top 40 for 2 years
5. Teaching and learning with generative AI:
policy contexts
2. Promoting the
design & use of AI for
inclusion and equity
4. Exploiting the
use of AI to support
futures of learning
AI and education:
Toward digital
humanism
1. Regulating
ethics of AI in
education
3. Building AI
competencies for
teachers and
students
7. AI: from moving targets
to proportionality
Production
and storage
of data
Access to
and control
of data
Dataand
algorithm
based
decision-
making
Human-AI
interfaces
and AI
devices
Human-
machine
collaboration
Individuals
Human
interaction
Sovereign
states
Environment
and
ecosystems
E.g. ChatGPT-3
▪ Data sources: Crawling
webpages (61.75%), Social
media (8.86%); Libraries(15.9%),
wikipedia (3.49%)
▪ Languages: EN (92.647%),
FR(1.819%), DE+ES +IT+PT+
NL (2.60%), …CN (0.099%)
▪ Unexplainablemethods
to generate outputs
▪ Generative AI doesn’t
understandsemantics
and the real world
1. Evolutionof
knowledge
production
2. Projection
social values
3. AI job skills;
challenging
learning
outcomes
Education: knowledge
evolution
1. A human-AI collaboration model
for examining AI in education
8. Discriminator
of computational skill
performance
Supercharger
of literature reviews
Transformer
across symbolic
representationsof human
thinking
Blinded
to values and the real
world
Human-
machine
collaboration
Human agency and
motivation
Human-
accountable
discovery and
interaction
Human protecting
linguisticand
cultural diversity
Greater harm for environment
and ecosystems
Generative AI is an AI insides,
insides of human thinking processes
New threat for the security of
digital learning infrastructure
9. Teaching and learning with …
ICTs
Generative AI
Predicative AI
Externalized info. processors
Internalized supercharger
for thinking → embodied
cognitive tools
Semi-internalizeddata advisor
in decision loops
Embodied
ethics of AI
+
AI society
citizenship
…… ……
11. Controversy 1:
Data deprivation worsening digital poverty
Internet
penetration
needed for
data
production
Digital
literacy for all
citizens to
ensure the
local
relevance and
cultural
diversity of
data
Affordability
of the
internet or
data
subscription
Local
availability of
AI developers
Accessibility
of AI chips
and
computing
Capabilities
of mining
data from
Internet
Funding and
technology
capabilities of
training
locally
relevant
models or
localizing
foundation
models
Domestic
governance
of the
development
and use of AI
A taxonomy for reducing AI divides
→ Define and govern the entire life cycle of the data-based productivity
12. No response or
N/A
137 countries
EU, US,
China
20+countries
Adjusting copyrightslaws
(labeling AI-generatedcontent)
Regulations ongenerative AI
Ethics of AI (including in
education)
Capacities for proper use of
generative AI ineducation
EU, US, China,
Japan?
Singapore
Reflections onimplications for
curriculumand assessment
General dataprotectionlaws
No response or N/A
70 countries
National strategies onAI
No response or N/A
No response or N/A
No response or N/A
No response or N/A
Tiered shortage of domestic regulations on AI particularly GeAI
Controversy 2:
Outpacing national regulatory adaptation
13. ▪ Violating GDPR: New York Times lawsuit on using its
content without consent
▪ Violating right to be forgotten (GenAI can’t unlearn)
Controversy 3:
Use of content without consent
14. The Foundation Model Transparency Index (Stanford University)
Examples of foundation models
Indicators
Llama2
(%)
ChatGPT-4
(%)
PaLM 2
(%)
Data 40 20 20
Datalabor 29 14 0
Compute 57 14 14
Methods for model’s training 75 50 75
Model basics 100 50 67
Model access 100 67 33
Capabilities 60 10 80
Risks 57 57 29
Model mitigations 60 60 40
Distribution 71 57 71
Usage policies 40 80 60
feedback 33 33 33
Impact 14 14% 0
Average scores 57 47 39
Controversy 4: Unexplainable AI models
used to generate outputs
→ Is the large-languagemodel a category mistake?
15. ▪ AIGC: Disinformation, misinformation, hate speeches
▪ Data polluted by AI lead to collapse of GenAI models
Controversy 5:
AI-generated content polluting the internet
16. GitHub’s Hallucination Leaderboard : the actual errors are worse
Model Accuracy Hallucination Rate
GPT 4 97.0 % 3.0 %
GPT 4 Turbo 97.0 % 3.0 %
GPT 3.5 Turbo 96.5 % 3.5 %
Llama 2 70B 94.9 % 5.1 %
Llama 2 7B 94.4 % 5.6 %
Llama 2 13B 94.1 % 5.9 %
Anthropic Claude 2 91.5 % 8.5 %
Google Palm 2 87.9 % 12.1 %
Google Palm 2 Chat 72.8 % 27.2 %
Controversy 6:
Lack of understanding of the real world
17. ▪ Data-poor populations have limited digital presence online →
Their voices are not represented in the data → GPT models can
further marginalize already disadvantaged
▪ LLMs may propagate race-based medicine
▪ ChatGPT replicates gender bias in recommendationletters
▪ Boston Consulting Group: The diversity of ideas from GPT-4 is 41%
lower human ideas.
Controversy 7:
Threatening plural knowledge construction
18. Nearly zero-cost generation
and sharing of deeper
deepfakes especially: taking
<25 mins and nearly $0 to
create a 60-second deepfake
pornographic video using just
one clear face image.
The totaldeepfake videos
by October 2023 presents
a 550% increase over 2019
with 98% being deepfake
pornographyand 99% of
targeting are women.
Controversy 8:
Generating deeper deepfakes
https://www.homesecurityheroes.com/state-of-deepfakes/#key-findings
20. 3.1 Safety of digital infrastructure and value projection
▪ Mandate the use of local data in training models
▪ Local EdGPT
21. 3.2 Age appropriateness of chat-based learning
for children
Age limitation for independent conversations with GenAI
platforms (13 years or 16 years)
22. 3.3 Pre-cooked content for learning
3.3.1 Content hallucinations lead to knowledge illusions
3.3.2 Undermining thinking and intellectual development
3.3.3 Inert innovation and intelligence enfeeblement
24. 4.1 Regulate generativeAI in education
▪ Risk categorization of AI for education: draw the red line
▪ Accountabilities of all key duty-bearers
▪ Validating AI for education to ensure ethical principles,
linguistic and curriculum relevance, cultural and age
appropriateness
▪ Promoting teacher and student agency, human motivation,
and intellectual development
25. 4.2 Build AI competencies for teachers and students
UNESCOAI competencyframework for teachers (under development)
Aspects Progression
Acquisition Deepening Creation
Human-centred
Mindset
Benefit-risk analysis Human accountability AI society
responsibility
Ethics of AI
Ethical principles Safe and responsible
uses
Co-creating AI ethical
rules
AI Foundations &
Applications
Basic AI technique and
applications
Applicationskills Creating with AI
AI Pedagogy AI-assisted teaching AI-pedagogy integration AI for pedagogical
transformation
AI for Professional
Development
AI enabling lifelong
professional learning
AI to enhance
organizationallearning
AI for professional
transformation
▪ the primary users of AI in education
▪ the designers and facilitators of students’ learning with AI
▪ ultimate safeguards for ethical uses across learning settings
▪ role models of lifelong learners about AI
Minimum
Standard
for teachers
What AI foundations and application skillsshould all teachers learn?
26. UNESCO AI competency framework for school students
(under development)
Competency Aspects
Progression Levels
Understand Apply Create
Human-centred mindset
Human Agency Human Accountability AI Society Citizenship
Ethics of AI
Embodied Ethics Safe and ResponsibleUse Ethics by Design
AI techniques and
applications AI Foundations Application Skills Creating AI Tools
*AI system design Problem Scoping Architecture Design Iteration and
Feedback Loops
▪ the responsible users of AI
▪ the active co-creators of AI
▪ the leaders of design and development of next generations of AI
Minimum standard:
AI literacy
What ethics of AI should we help students to learn and practice?
27. To launch
▪ AI competency framework for school teachers
▪ AI competency framework for school students
28. A “Human-agent and pedagogically proper interactions” template
Potential
but
unproven
uses
Appropriate
domains of
knowledge
or problems
Expected
outcomes
Appropriate
GenAI tools
and
comparative
advantages
Requirements
for the users
Required
human
pedagogical
methodologies
and example
prompts
Possible
risks
AI Advisors
for
augmented
research
outlines
……
4.3 Guide instructional design on uses of GenAI
29. 4.4 Trade-offs and tests of use modes of
Generative AI in teaching and learning
• Local(ized) EdGPT for digital textbooks(eg. Republic of Korea)
• Transformers for curriculum co-design (eg. Oak National Academy)
• Formative discriminators for computationalskills (e.g. auto-
coaching learning of coding )
• Superchargers for higher-order thinking or inquiry-based learning
Use petrol engines as indicators to test the electric motors?
What sort of tests can lead to the leap innovation in pedagogy?
Does the leap innovation in pedagogy really exist?
Trade-off: Why should we use? When to use?
30. What sort of human-digital/AI social contract
should we build?
31. A fundamentalask:
Long-term implications of GenAI for education?
Post-AI digital humanity and learning:
Co-existing with human by design
Human-AI-hybrid intensive learning?