1 of 101

<< Introduction to Machine Learning >>

Week 6

CCAI 9012

GenAI Solutions to Global Challenges: �Using AI Creatively and Responsibly

Kam-Ming Mark Tam

Class Coordinator

Impressionist-style scene of human–AI co-existence

Generated with ChatGPT (OpenAI), 2026.

Kanxuan HE

Naixiang GAO

Yifan XIE

Chao TANG

May Loaay Mohamed EL-HADIDI

Joseph Jing Hymn WONG

Christina Ka Man CHU

Charles Wai Lam TO

Shum Nga MAN

BT 1 (ARCH 2056/7330) Building Technology 1 : Building Principles

08/03/2026

1

Kam-Ming Mark TAM

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

1

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

2 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

2

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

3 of 101

Review of Timeline

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

3

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

4 of 101

Milestones & Corresponding Items

A.1.1: Responsibility and AI

A.1.2: Mechanics of AI

A.1.3: Datasets in AI

LAST NIGHT

2026.04.01

2026.04.29

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

4

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

5 of 101

Milestones & Corresponding Items

Team Formation Form

Mid-term Pecha-Kucha Presentation

Crit-Style Peer Review (at mid-term & finals)

Final Presentation Submission

Individual Reflection

Peer Assessment

2026.02.25

2026.03.18

by tutorial sign-up

2026.04.22

2026.05.06

2026.05.06

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

5

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

6 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

6

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

7 of 101

Machine Learning as Useful Mapping

IN

ℳ

OUT

x

y

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

7

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

8 of 101

Machine Learning as Useful Mapping

IN

ℳ

OUT

design of our structure

structural performance

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

8

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

9 of 101

Machine Learning as Useful Mapping

IN

ℳ

OUT

what we eat

how healthy are we

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

9

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

10 of 101

Data Points

IN

ℳ

OUT

x

y

(x1, y1),

(x2, y2),

…

(xN, yN)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

10

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

11 of 101

Data Points

IN

ℳ

OUT

x

y

(1.0, 1.0),

(1.5, 2.8),

…

(3.8, 5.4)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

11

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

12 of 101

Data Points Visualised

ℳ

x

y

IN

OUT

x

y

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

12

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

13 of 101

ℳ

x

y

IN

x

y

OUT

IN

x

y

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

13

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

14 of 101

Linear Regression

y=wx+b

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

14

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

15 of 101

Model Training

x

y

y=wx+b

Model learns parameters

w

b

1

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

15

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

16 of 101

Mentimeter: Features Brainstorming

IN

ℳ

OUT

?

y

ACTIVITY

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

16

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

17 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

17

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

18 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

18

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

19 of 101

Example Tasks

Overall Goal

Mapping

Dataset

SUPERVISED

—Regression

—Classification

Pattern Recognition

f: X → y

X, y

UNSUPERVISED

—Clustering

—Dimensionality Reduction

Knowledge Disccovery

f: X → z

X

z: latent factors

REINFORCEMENT

... gaming, trading, layout editing, autonomous vehicle, robot control…

Sequential Decision Making

f: X(t) → X(t+1)

X(t), a(t), r(t), X(t+1)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

19

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

20 of 101

Example Tasks

Overall Goal

Mapping

Dataset

SUPERVISED

—Regression

—Classification

Pattern Recognition

f: X → y

X, y

UNSUPERVISED

—Clustering

—Dimensionality Reduction

Knowledge Disccovery

f: X → z

X

z: latent factors

REINFORCEMENT

... gaming, trading, layout editing, autonomous vehicle, robot control…

Sequential Decision Making

f: X(t) → X(t+1)

X(t), a(t), r(t), X(t+1)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

20

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

21 of 101

What is the differences between supervised & unsupervised learning?

Labels or No Labels

IN

ℳ

OUT

x

y

(x1, y1),

(x2, y2),

…

(xN, yN)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

21

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

22 of 101

x

y

ℳ

x

y

IN

x

y

OUT

IN

x

y

→

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

22

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

23 of 101

Classification

Supervised Learning

x1, x2…

class

0: red�1: blue

x1

x2

→

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

23

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

24 of 101

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

24

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

25 of 101

https://nanonets.com/blog/how-to-do-semantic-segmentation-using-deep-learning/

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

25

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

26 of 101

Example Tasks

Overall Goal

Mapping

Dataset

SUPERVISED

—Regression

—Classification

Pattern Recognition

f: X → y

X, y

UNSUPERVISED

—Clustering

—Dimensionality Reduction

Knowledge Disccovery

f: X → z

X

z: latent factors

REINFORCEMENT

... gaming, trading, layout editing, autonomous vehicle, robot control…

Sequential Decision Making

f: X(t) → X(t+1)

X(t), a(t), r(t), X(t+1)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

26

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

27 of 101

Clustering

Unsupervised Learning

x1,x2

cluster

discovered by algorithm

→

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

27

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

28 of 101

Clustering

Unsupervised Learning

x1,x2

z1,z2

[1,0] for points in Cluster 1

[0,1] for points in Cluster 2

algorithm determines what groups are

→

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

28

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

29 of 101

Clustering

Unsupervised Learning

x1,x2

z1,z2

more different

more similar

[1,0] for points in Cluster 1

[0,1] for points in Cluster 2

algorithm determines what groups are

→

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

29

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

30 of 101

Dimensionality Reduction by Manifold Learning : Concepts

H = W = 28

​

X ∈ ℝ H × W = 784

0

H

W

x1

x784

…

+

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

30

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

31 of 101

Dimensionality Reduction by Manifold Learning : Concepts

x1

x784

…

+

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

31

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

32 of 101

Example Tasks

Overall Goal

Mapping

Dataset

SUPERVISED

—Regression

—Classification

Pattern Recognition

f: X → y

X, y

UNSUPERVISED

—Clustering

—Dimensionality Reduction

Knowledge Disccovery

f: X → z

X

z: latent factors

REINFORCEMENT

... gaming, trading, layout editing, autonomous vehicle, robot control…

Sequential Decision Making

f: X(t) → X(t+1)

X(t), a(t), r(t), X(t+1)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

32

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

33 of 101

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

33

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

34 of 101

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

34

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

35 of 101

Modelling Decision Sequences

A Graph-Based Grammar for Structural Design Using Deep Reinforcement Learning (2024), Bleker, Tam & D’Acunto

  • AlphaGo

Playing Chess

Designing Layout

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

35

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

36 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

36

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

37 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

37

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

38 of 101

Encoding Data

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

38

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

39 of 101

Data Points

1.0

1.0

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

39

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

40 of 101

Data Points

(1.0, 2.0)

1.0

2.0

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

40

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

41 of 101

Motivation: Modelling Diverse Non-Euclidean Designs

x1

x2

design distances

0.0

0.0

0.0

1.0

1.0

-1.0

0.0

-1.0

-1.0

-1.0

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

41

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

42 of 101

Motivation: Modelling Diverse Non-Euclidean Designs

x1

x2

distance

vector spaces allow us to represent data geometrically, making it possible to interpret similarity and difference in terms of distance.

0.0

0.0

0.0

1.0

1.0

-1.0

0.0

-1.0

-1.0

-1.0

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

42

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

43 of 101

Data Points

(x1,…, xD)

…

D

1

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

43

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

44 of 101

Multiple Data Points

(x1,1,…, x1,D)

(x2,1,…, x2,D)

…

(xN,1,…, xN,D)

…

…

…

…

D

1

1

N

x1

x2

​

xD

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

44

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

45 of 101

Image as Data Matrices

https://ai.stanford.edu/~syyeung/cvweb/tutorial1.html

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

45

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

46 of 101

Feature Engineering

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

46

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

47 of 101

Motivation: Modelling Diverse Non-Euclidean Designs

x2

x1

x1

x2

x1

x2

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

47

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

48 of 101

Data Points

(0.0, 0.0)

0.0

0.0

x2

x1

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

48

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

49 of 101

Example Features: Iris Dataset

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

49

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

50 of 101

Feature Engineering: One-Hot Encoding

cat

1.0

0.0

 

dog

0.0

1.0

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

50

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

51 of 101

Representation Learning

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

51

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

52 of 101

Modelling Diverse Non-Euclidean Designs

?

?

?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

52

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

53 of 101

Text as Embedding Vectors

A long winded sentence…

…

ℳ

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

53

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

54 of 101

Embedding Space Exploration

Embedding Projector : https://projector.tensorflow.org/

ACTIVITY

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

54

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

55 of 101

Objectives & Optimisation

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

55

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

56 of 101

Optimisation

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

56

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

57 of 101

We want models to be accurate…

We want to reduce error….

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

57

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

58 of 101

Training =

optimising model parameters to improve prediction accuracy

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

58

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

59 of 101

ℳ

x

y

IN

x

y

OUT

IN

x

y

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

59

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

60 of 101

x

y

x

y

or

?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

60

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

61 of 101

loss = error0 + error1 +…+ error5

loss = error0 + error1 +… + error5

x

y

x

y

or

error0

error6

error0

error6

?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

61

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

62 of 101

Model Training

x

y

The training process involving iteratively updating the parameters that define a model…

y=wx+b

w

b

1

 

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

62

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

63 of 101

Data Generalisation

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

63

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

64 of 101

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

64

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

65 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

65

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

66 of 101

Machine Learning Life Cycle

  1. Gathering Data – Collect relevant raw data from various sources such as databases, sensors, APIs, or web scraping.
  2. Data Preparation – Clean and organize the data, removing duplicates and handling missing or inconsistent values.
  3. Data Wrangling – Transform and structure the data into a usable format suitable for analysis and model training.
  4. Analyse Data – Explore the data through visualization and statistical analysis to identify patterns and insights. Train
  5. Train Model – Use the processed data to train machine learning algorithms, allowing the model to learn from examples.
  6. Test Model – Evaluate the model’s performance using unseen data to measure its accuracy, precision, and generalization ability.
  7. Deployment – Integrate the trained model into production systems where it can make real-world predictions and be monitored for performance.
  • Machine Learning
  • Life Cycle

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

66

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

67 of 101

Data Preparation

x

y

x

y

or

?

x

y

😈

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

67

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

68 of 101

Data Wrangling ↔ Analyse Data

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

68

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

69 of 101

Model Selection

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

69

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

70 of 101

Real World Data

Real Relationship

Building model from limited realife data subset…

Training Data Scenario 1

Training Data Scenario 2

Training Data Scenario 3

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

70

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

71 of 101

Dtest

Dataset D

Dtrain

train on

Dtrain

test on

Dtest

ℳ

✂️

🫣

👁️

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

71

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

72 of 101

Validation via Dtest

Real-Life Data

Acquired Dataset

split into Dtrain and Dtest

​

​

Training via Dtrain

Model 1

Model 2

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

72

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

73 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

73

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

74 of 101

Number → Number

Image → Categories

Image → Number

Text → Text

Text → Image

Image → Text

…

​

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

74

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

75 of 101

Become a process thinker…

Build complexity from simple parts…

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

75

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

76 of 101

Singapore University of Technology & Design, Introduction to Design (2018)

Start drawing these flow diagrams!

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

76

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

77 of 101

ML as Versatile Data Mapping

IN

ℳ

OUT

x

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

77

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

78 of 101

Machine Learning as Useful Mapping

IN

ℳ

OUT

text

text

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

78

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

79 of 101

Machine Learning as Useful Mapping

IN

ℳ

OUT

9.5

Analyse this text/image under criteria A, B, C…

× 100

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

79

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

80 of 101

AI as Social Interface

Problem Solving with AI

Source: Student work from design studios of Prof. Christian J. Lange (2024)

Martin Dun Man LAP

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

80

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

81 of 101

Iteration & Collaboration

AI Application Principles

Source: Generated with MidJourney, 2025.

/imagine A bridge spanning across 100 meters connects two separated cliffs, which are partially covered with lush vegetation at their bases, while their upper portions reveal bare, rugged rock formations. Below the bridge lies a vibrant city, bustling beneath this architectural marvel. The structure features expressive steel elements painted in white, with a truss layout that is organic in form, exhibiting a variety of shapes and curves. The design is thoughtfully optimized in sizing, incorporating decorative elements that evoke the distinctive style of Calatrava. Crafted in the 20th century, the bridge showcases an extremely realistic appearance when viewed from an elevation perspective, presenting a frontal view that highlights its elegant, innovative, and dynamic aesthetic.

Class elective work of Marco Ma Kwun Ho (2024)

Show as how you engage with AI…

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

81

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

82 of 101

Example Course Project

Example Course Project

<<Make Happy Art for Hospitals>>

Natalie Ka Sin FONG

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

82

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

83 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

83

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

84 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

84

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

85 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

85

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

86 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

86

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

87 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

87

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

88 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

88

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

89 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

89

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

90 of 101

Natalie Ka Sin FONG (2024)

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

90

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

91 of 101

Mentimeter: Algorithmic Process Brainstorming

ACTIVITY

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

91

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

92 of 101

Announcements + Logistics

Modelling Preliminaries

One Interpretation of ML

Main Classifications

[Break]

Fundamental Concepts

Core Workflow

Process Thinking

Course Project

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

92

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

93 of 101

Course Project Check List

  • What problem?
  • For whom?
  • Have you looked for precedents?
  • What small test will you run?
  • Is it achievable within 5 weeks?
  • How will you know if it helped?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

93

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

94 of 101

Current Course Project Statements

1️⃣ Generative AI for Public Awareness Posters

Public awareness campaigns exist on campus and throughout the city. However, they are limited by long production cycles, design constraints, and inconsistent messaging to properly capture attention to urgent and current issues.

Thus, this project aims to explore how generative AI can be used to create posters that are both content-specific and engaging to the public, and that can be effectively used across campus and city spaces.

The goal can be reached by doing AI analysis and evaluation of existing datasets and information, as well as applying knowledge of generative methods.

​

2️⃣ LLMs and Politically Sensitive Responses

Large language models are increasingly used in everyday life and serve as a primary tool to gather information for many users. One of these many uses involves LLM responses to politically sensitive questions.

While their neutrality is often assumed due to safety guidelines and bias management claimed by the companies that develop such models, LLM responses ultimately vary since outputs are shaped by variables such as training data and alignment policies.

Such differences in response may lead AI systems to unknowingly influence user perception.�

3️⃣ AI-Based Solution for Older Adults Using Smart Devices

In recent years, Hong Kong has transformed into a web-based society, and older adults (60 years old and above) must learn to use smart devices and access the internet for daily life and entertainment.

However, many older adults suffer from cognitive impairments, including dementia, which affects memory and comprehension. Additionally, worsened physical abilities such as chronic pain — affecting 37.1% of the local population aged 60 and above — depreciate the user experience of using smart devices.

These factors make it difficult for older adults to navigate digital systems and remain connected in society. Therefore, an AI-based solution, such as a mobile application, can be developed to help older adults learn to use smart devices more effectively.�

4️⃣ AI Optimisation for Urban Planning

Problem Statement: Hong Kong has become one of the world's most populated cities. Globally, urban agglomeration has caused cities to become cramped and harder to live in. At the current pace, this risks the quality of life of future generations living in cities.

The over-use of facilities like public transportation, and their inadequate distribution, has exacerbated this issue of fast-growing populations.

Possible Direction: Utilize AI-based algorithms to optimize urban planning and design, such that space is maximized while ensuring green and liveable environments. These algorithms can assist urban planning by optimizing the distribution of public services, facilities, and transportation. Feeding this data into LLMs, findings can be summarised into easy-to-understand plans of action for governmental bodies and other platforms.�

5️⃣ AI for Anti-Scamming

Recently, scamming has been spread across the whole Asia. For example, in Japan, the high-trust society leads into trust with banks and police. So victims usually easily mistrust the scammers.

Scamming not only erodes public trust, but also causes huge financial losses. Therefore, our team is going to use AI to help solve the scamming issue.

Compared with traditional anti-scam algorithms, AI can deal with multiple variables and thus can be used in more complex situations for anti-scamming.

​

6️⃣ AI Image Detection / Authenticity Awareness

With the help of AI image generation tools, AI-generated images have penetrated every corner of information dissemination. From KOL content on social media to news captions, it is becoming increasingly difficult for people to distinguish what is real.

During Hurricane Helene (September 24–29, 2024), purportedly AI-generated images circulated on social media, reportedly misleading the public and hindering disaster response efforts.

When any “photo” can be forged, the authority of visual evidence is weakened. People begin to doubt visual information and must invest extra effort to determine authenticity.

The core question of this project is:

How can we design an AI image detection tool that helps users recognize AI-generated content, understand detection logic, and remain aware of the tool’s limitations and ethical responsibilities?

The emphasis is not on achieving 100% accuracy, but on exploring ethical dilemmas and social impact.

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

94

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

95 of 101

No Baseline for Comparison | No Explicit Evaluation Layer

  • Problems are framed at societal or continental scale:
    • “Across Asia”
    • “Hong Kong urban future”
    • “Public trust erosion”
    • “Information dissemination everywhere”
  • But no:
    • District
    • Dataset
    • User group
    • Defined context
  • This makes them...
    • rhetorically grand
    • operationally vague
    • Intractable to address…

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

95

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

96 of 101

No Clear Input–Output Mapping

The system’s process is not described.

​

  • Examples:�“AI-based mobile application.”�“Image detection tool.”

​

What goes in?�What comes out?

​

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

96

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

97 of 101

No Baseline for Comparison

Claims imply improvement, but no comparison is specified.

​

  • Examples:�“More effective than traditional methods.”�“Optimise facility distribution.”

​

  • Compared to what?
  • What to compare?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

97

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

98 of 101

No Baseline for Comparison | No Explicit Evaluation Layer

Claims imply improvement, but no comparison is specified.

​

  • Examples:�“More effective than traditional methods.”�“Optimise facility distribution.”

​

  • Compared to what?
  • What to compare?

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

98

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

99 of 101

Assumption Without Evidence

​

  • Some statements assume:
    • AI is more complex and therefore better
    • AI is neutral unless proven biased
    • Infrastructure is overused
    • Detection is possible and desirable
  • But do not:
    • Define how those assumptions are validated
    • Identify competing explanation

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

99

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

100 of 101

Review of Time for Course Project Development

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

100

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU

101 of 101

Thank You for your attention!

CCAI 9012 : GenAI Solutions to Global Challenges: Using AI Creatively and Responsibly

2026.02.25

101

Kam-Ming Mark TAM | Hongshan GUO | Waishan QIU