MSBA7033 — Generative Artificial Intelligence

Class 1:
Course and Topic
Introduction

Prof. Dawei Wang

Email: [email protected]

1
Prompt: person exiting a door, entering into a forest and mountains | Tool: Adobe Firefly
Class Agenda
Introductory class

1. Get to know the course

2. Get to know the topic

3. Get to know me

4. Get to know you

5. Get to know the materials

6. Get to know the project

2
ABOUT THE COURSE

Course Topic About me About you Materials Project

3
Classroom Format

Universities are factory lines, Universities are places of self-

!
they manufacture degrees discovery and experimentation

Students must fit into a box and Students should think creatively,
!
think homogeneously independently and globally

Students are products and Every student is unique and

!
teachers are manufacturers teachers facilitate their learning

4
Classroom Rules

! You should not take photo or record

" You should not talk during lectures

# You should participate actively and ask questions

$ You should not directly generate answers using AI

5
Exciting Things in This Course

! In-depth knowledge of the field

" Tiny Fish guest speaker

# “Beat the Bot” exercise

$ AI-based assessments

% Six business cases

16
Class Schedule
No. Date Topic Deadlines Class Activities
1 28 Nov 2024 Introduction - -

2 02 Dec 2024 Intelligence - Beat-The-Bot game

3 05 Dec 2024 Dev: Data - -

4 09 Dec 2024 Dev: Software - AI-based assessments

5 12 Dec 2024 Dev: Hardware 20-min quiz Nvidia case

6 16 Dec 2024 App: Text Group proposal Google case

7 19 Dec 2024 App: Image - Adobe case

8 02 Jan 2025 App: Video 20-min quiz Rayvision case

9 06 Jan 2025 App: Code Assignment Tiny Fish case

10 09 Jan 2025 ESG 20-min quiz OpenAI case

11 13 Jan 2025 Group presentation Final report -

17
Grading Format
Participation, assignments + project | No midterm and final examination

Assessment methods Time period or deadlines Weight in percentage

Class attendance From Class 2 to Class 10 10%

In-class quizzes From Class 2 to Class 10 30%

Individual assignment On Class 4 10%

Group project registration On Class 3 -

Group project proposal On Class 5 10%

Group project presentation On Finals Week 20%

Group project report On Finals Week 20%

Total: 100%

18
Individual Assignment (Release on Class 6 | Due on Class 9)
You are a consultant for a recruiting company, researching a new market for the client to go into.

Q1: Choose a geography: China Mainland, China Hong Kong or USA.

Q2a: How big is the market?

Q2b: How big is the AI-recruiting market?

Q3a: In the three stages of recruitment, what are the existing companies, solutions and prices

Q3b: Discuss in terms of natural, rule-based and AI-based intelligence

Q4: Choose one AI-based solution, what’s the data, software and hardware?

Q5: Which market is the most promising?

19
ABOUT THE TOPIC

Course Topic About me About you Materials Project

20
Generative AI

21
Generative AI Sees the Fastest User Adoption
ChatGPT reached 1 million users in 5 days, the fastest technology adoption in the history of IT

Number of months to reach 100 million users

300

225
216
Number of Months

150 132

75 54 60
30
18
2 9
0
ChatGPT TikTok Youtube Instagram Facebook Twitter Spotify Netflix

Source: UBS and Yahoo Finance

Note: Threads is not counted because most of its users came from Instagram.
22
All Big Techs Are Pursuing Generative AI
Big-tech companies are pursuing a large-language model (LLM) arms race

Company Large-language model Number of parameters Chatbot

Microsoft / OpenAI GPT 3.5 175 B ChatGPT

Microsoft / OpenAI GPT 4.0 1.7 T ChatGPT

Google Gemini >1.7 T (rumored) Bard

Google / Anthropic Claude 2 137 B Claude

Meta Llama 2 70 B -

Meta OPT-175B 175 B Blenderbot

Baidu Ernie 4.0 >1 T (rumored) Ernie Bot

Moonshot AI K0-math 200 B Kimi

- Human brain ~90 B -

23
Applications of Generative AI across Modalities
Modality Applications Examples

Text Content writing Generating personalised emails; Service chatbots

Code Code generation Accelerating coding and programming

Image Stock image, image editor Social media marketing, creating presentation slides

Voice Text to voice, translation Creating professional voice-over

Music Music generation Synthesising new music in a particular style or genre

3-D 3-D object generation Generating 3-D objects in video games

Video Short video creation, face-swap Swapping faces in videos, animating photographs

Source: Adapted from McKinsey & Co’s report on Exploring opportunities in the generative AI value chain
24
Artificial Intelligence Stages
AI ANI AGI ASI
Stages of AI
Artificial intelligence Artificial narrow intelligence Artificial general intelligence Artificial super intelligence

Computer system that manages Computer system that Computer system that is as Computer system that is much
Description
to do a task but not effective specialises in one area smart as an average human smarter than the best human

Could be much BETTER

Autonomous /
than humans in ANY
Adaptive
task available

Could be an assistant
Augmented that does ANY tasks as
good as humans

Could beat humans in

Predictive /
SPECIFIC tasks such as
Generative
Go and computer games

Could do some
Repetitive repetitive tasks as good
as humans
Timeline 1950s 2000s 2030s (?) 2060s (?)
25
ABOUT ME

Course Topic About me About you Materials Project

26
Industry Practice and Management Research
Experience Year Department / Area Role / Position

Banking experiences

Mercer 2012 Human resource consulting Information Product Solutions

Credit Suisse 2013 Corporate and institutional banking Corporate and Institutional Clients

Research experiences

2014 Bachelors in Business Majoring in management

NUS Business
School
2016 Research assistant Focusing on social psychology

2019 Masters in Management and Organizations Majoring in management

Northwestern
Kellogg
2022 PhD in Management and Organizations Majoring in artificial intelligence

Current academic positions

HKU Business Innovation and Information Management

2022—present Assistant professor
school Management and Strategy
Northwestern
2024—present External faculty Artificial Intelligence and Creativity
NICO
27
About Our Teaching Assistant

Tutor: Zikang Dai (Ken)

Email: [email protected]

Consultation times: By appointment

Background:

Bachelors in Finance, Monash University

Masters in Global Management (MGM 2024), HKU

Full-time RA and TA at HKU

Applying to PhD at HKU

ABOUT YOU

Course Topic About me About you Materials Project

31
ABOUT THE COURSE

Course Topic About me About you Materials Project

33
My Academic Role Model

Herbert A. Simon (born 1916, died 2001)

BA and PhD in Political Science

The University of Chicago

Professor in Administration
Carnegie Mellon University

APA Award for contributions to psychology (1969)

Turing Award for contributions to computer science (1975)
Nobel Prize Laureate for contributions to economics (1978)

He helped found the first department in computer science

34
Where This Course Comes from?
Books by Herbert A. Simon recommended for this course

Simon (1947) Simon (1969)

Administrative The Sciences of the
Behavior Artificial

Question: Why use two books that were published

half a century ago to talk about recent and future
developments in generative AI?

Answer: Because these books discuss the most

important question of the century and beyond:
what is and should be the relationship between
humans and machines? This question is the central
theme of this course and my research.

35
The Real Beginnings of (Generative) AI

The IBM 650 was Carnegie Institute of Allen Newell and General Problem Solver
released in year 1954. Technology. Herbert Simon. at Rand Corp.

The IBM 650 was a huge The computer was They met at the Rand One of the earliest
computer. However, housed at the basement Corp. which is a US artificial intelligence
Herbert Simon of the business school. military research systems was developed
persuaded his university Computer science was institution. The institute here, by Simon and
to purchase the first studied under the funds many military- Newell. It was used to
computer in any business school. Later it related research assist decisions in missile
university. became its separate including artificial defense.
degree. intelligence.

36
Herbert Simon’s Theory on Artificial Intelligence
The difference of this course to one that is offered computer science departments

Logical rationality Ecological rationality

Computer science Information systems

Vacuum Environment
System System

Component B Component B

Component A
VS Component A

Component C Component C

Component D Component D

Component E Component E

37
What Herbert Simon Would Say about This Course?

What computer scientists are saying now What the founder of computer science would say

Focus on models or algorithms Focus on data, people and systems

Design of models or algorithms Design of inter-dependent systems

Techniques of programming models Applying models and seeking feedbacks

Micro-perspective and internal logics Macro-perspective and implications

Assume people will use generative AI Whether, how people actually use generative AI

How computers can replace humans How computers can augment humans

38
Antithesis and Objectives of This Course

What this course is NOT What this course IS

How to program or code generative AI A working, general theory on generative AI

How to use specific generative AI models How to evaluate specific generative AI models

Prefer mathematical models and algorithms Comparison to human and rule-based models

Which model perform better than others General strengths and limitations of generative AI

How to design algorithms and models How to design systems in organizations

Short-term gains instead of long-term gains ESG considerations in generative AI

39
Perspectives of This Course

Students will approach this class from three

perspectives:
• Theory
Concepts • Applications
• Implications

Students will be capable of engaging in

related careers:
Applications Implications • Entrepreneurs
• System designers
• Policy-makers
• Consultants

40
GROUP PROJECT
INTRODUCTION

Course Topic About me About you Materials Project

41
Generative AI Represents The Hottest Startup Area
According to CB Insights, Generative AI startups raised $10 billion in capital in 2023; Investors
even begged Midjourney to take their money but the company did not require any!

Seed funding acquired by company Average seed funding in silicon valley

120
US$ 101 million

90

60 US$ 55 million

30

US$ 8 million
US$ 1 million
0
Theranos FTX Pika AI Stability AI

Source: Crunchbase (https://www.crunchbase.com/)

42
Generative AI Will Be the Biggest Growth Driver
McKinsey & Co. predicts that generative AI has the potential to generate $2.6 trillion to $4.4
trillion each year. In comparison, UK’s GDP in 2021 was $3.1 trillion.

Total % of Total $ amount Marketing Customer Product, Software Supply Risk and Strategy & Talent and
Industry and sales operations R&D engineering chain legal finance
Corporate IT
organization
industry revenue in billion(s)

High Tech 4.8-9.3 240-460

Banking 2.8-4.7 200-340

Education 2.2-4.0 120-230

Healthcare 1.8-3.2 150-260

Insurance 1.8-2.8 50-70

Consumer 1.4-2.3 160-270

Adapted from Exhibit 4 in The economic potential of generative AI: The next productivity frontier Low impact High impact
43
 Imagine you are an innovative team of

entrepreneurs
starting a business in generative AI..

44
How Startups Grow
Funding amount, shareholdings, valuations according to funding and operating stages

Funding round Pre-seed Seed Series A Series B+ IPO

Stage Idea Pre-launch Growth Expansion Public

Friends, family, Angel investors,

Bank funds,
Investors grants, incubators, venture capitalists, Venture capitalists Venture capitalists
IPO investors
crowdfunding crowdfunding

Money raised US$10k-50k US$500k-2m US$2m-15m US$30m-60 m US$30m-100m

Founders’ shares 90% 70% 60% 45% 35%

Valuations US$100k-500k US$3m-6m US$10m-15m US$70m-100m US$1b

45
 Your task is to prepare a pre-seed

investor pitch and report

that can potentially be used to procure
fundings from angel investors

46
Potential Startup Ideas
Type Modality Applications Examples

Text Content writing Stock predictor or picker, sentiment analyzer

Single-modal
Image Stock image, image editor Professional photograph generator
applications
Music Music generation Generative-AI based DJ

Text and video AI based assessments App that gamify recruitment assessments
Multi-modal
Video AI video interview Software that can help HR automate interviews
applications
Audio and text AI meeting assistant App that moderates and facilitates meetings

Data Training data crawler Service that crawls specialised training data

Support Design Human-centered design Service that put human first in AI applications

ESG Measure ESG of AI startups Service that evaluates AI startups’ ESG

47
Sample Investor Pitch Deck Contents
1 The problem What’s the problem you are trying to solve?

2 Existing solutions What are existing solutions and their limitations?

3 The solution How do you solve this problem?

4 Market size How big is the pie?

5 Competitors Who will share this pie?

6 Target audience Who are your target audiences?
7 Go-to-market How to reach target audiences?
8 Traction What will be an estimate of your growth?

9 Business model Pricing and revenue strategies?

10 Team Who are you? Why should investors trust you?

11 Ask and use of funds How much are you asking and why?

12 Gantt chart What’s the targeted timeline and milestone?

13 Summary slide What is the take-away of your investor pitch?

Your project in this class should potentially help you win the next Jumpstarter

49
Q&A

50