CH 01 Eng S v1.0

Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

AI Education Curriculum for Junior Secondary Schools in Hong Kong

TEACHER VERSION AI Education Curriculum for Junior Secondary Schools in Hong Kong

RESOURCES

Chapter

1
Introduction to
Artificial Intelligence (AI)

Contents
1.1  1-2
1.1.1 Basic concepts of AI 10 1-4
1.1.2 AI applications and non-AI applications 15 1-5
1.1.3 Examples of AI failure 20 1-7

1.2  1-10


1.2.1 Definition and history of AI 25 1-11
1.2.2 Components of AI 65 1-12

1.3  1-20


1.3.1 Ethical considerations of AI 10 1-21
1.3.2 How AI applications affect our daily lives 35 1-21
Chapter 1: Introduction to Artificial Intelligence (AI)

1.1 

Learning Objectives


  
       

Today is Kelly’s birthday. Her parents have a big surprise for her.

1 2

Happy birthday, Kelly!


Now close your eyes, make a
wish and guess what birthday
Thank you Mum and Dad! present you’re getting.

Kelly closes her eyes, and her dad puts a Kelly opens the box, and a cute robotic
3 4
big box in front of her. pet dog jumps out.

Open the box now sweetie,


but don’t scream.

1-2
AI Education Curriculum for Junior Secondary Schools in Hong Kong

5 6
Happy birthday Kelly! Nice
WOW! It looks so cute! to meet you.

You can speak!?

7 8
This is not a real dog. It is a robot Artificial intelligence, AI.
dog with high intelligence.
AI? It sounds interesting.
What is AI?
High intelligence?
What do you mean?

It is interesting!
Let me explain AI to you.

1-3
Chapter 1: Introduction to Artificial Intelligence (AI)

1.1.1 Basic concepts of AI 10 BASIC


   
           
             

Example 1 (Basic) Example 2 (Intermediate) Example 3 (Advanced)

VIDEO VIDEO VIDEO

             


                  
           
   

           
         
        

1-4
AI Education Curriculum for Junior Secondary Schools in Hong Kong

1.1.2 AI applications and non-AI applications 15 BASIC


                  
                       

     
         

      ★

             

­€ €‚ƒ  



„✓/✗


1-5
Chapter 1: Introduction to Artificial Intelligence (AI)

    ★★

     

       


     





 

 

 

                    
   
                  
          ­     
 ­          


1-6
AI Education Curriculum for Junior Secondary Schools in Hong Kong

1.1.3 Examples of AI failure 20 BASIC


         
                 
         



              


          

 VIDEO

    ­                    


            

€      ‚       


 

1-7
Chapter 1: Introduction to Artificial Intelligence (AI)

        ★★

        ­           
­ ­   €‚

        

  ­

  €

  

  

 ƒ  ­   


 



      



         
       
      

1-8
AI Education Curriculum for Junior Secondary Schools in Hong Kong

1 2
Thank you, Mum. I have a basic I think AI is too powerful and
idea of what AI is, but now I am will replace humans one day.
a little scared.

That’s why you need to learn


What are you scared of? more about the possible
challenges of AI.

 Optional

       

______________________________________________________________________

______________________________________________________________________

              
  

  







1-9
Chapter 1: Introduction to Artificial Intelligence (AI)

1.2 

Learning Objectives

  
  
     
 


1 2
AI Dog, look at me, What are you guys doing?
I am Grandma.
Grandma...
We are training AI Dog
to recognise us.
AI Dog, look at me,
I am Brother. Brother...
That’s funny! AI Dog can
be trained?

3 4
Real dogs can recognise their owners Machines can learn?
naturally, but AI dogs need training!

Yes, machines can learn.


But how?
Let me explain.

Kelly, you just need to ask AI Dog to


look at you and tell him you are Kelly.
That’s called ‘machine learning’.

1-10
AI Education Curriculum for Junior Secondary Schools in Hong Kong

1.2.1 Definition and history of AI 25 BASIC




                    
       ­   ­
­ ­     ­   
  €       € 

‚ ƒ ­ ­      VIDEO


 ­ ­  ­„­
    ­      ­       ­  
 ­        ­         ­   
 ­  ­ ­†  ­        ‡   
­             ­     
 ­     ­      
      ­  ­  
   €€    ­  ­ 

„ˆ        ­     ­   ­  ­   
­  ­   ­   ­ ­        ‰ € 
€ ­ ­ €         

  

Š­‹        ­ ­    


  ‚ €       ­ 
 ­ €        ­ Œ
€     ­    €­ ­  
   ­  €            ‚   
     € €

1-11
Chapter 1: Introduction to Artificial Intelligence (AI)

History of AI

1950 1960 2010


Birth of AI Neural Network Big Data, Cloud Computing and Deep Learning

   ­ 


      

    

Simulate   


Think & Create
 



 AI Reasoning
See

AI




 



     
      



       

Hear


 €        Read & Write ­‚   

Speak Rabbit

   


         
HAPPY

      READ   


  

­  

  
   



1.2.2 Components of AI 65 INTERMEDIATE

                
                       
           
             
                        
          
          
       

provide information
Big Data Speech

Machine
AI Prediction /
algorithm Vision
Learning Generation

Cloud processing power Reasoning


Computing
  

...
1-12
AI Education Curriculum for Junior Secondary Schools in Hong Kong

                   
           
          
     

             
     
      
                  
       
                  
        
      

 ­  

     ★

€  
­‚


_________________________________________

_________________________________________



_________________________________________

ƒ ­‚ _________________________________________



_______________________________________________________________________

1-13
Chapter 1: Introduction to Artificial Intelligence (AI)

     ★★

          


    

 

______________________________________________

______________________________________________

 

    ______________________________________________

______________________________________________

   

_______________________________________________________________________

_______________________________________________________________________

  

    ­  


    
­  €
   ­
    ­   

1-14
AI Education Curriculum for Junior Secondary Schools in Hong Kong

    ★★

      


                   


 : 










1-15
Chapter 1: Introduction to Artificial Intelligence (AI)

      ★★★

         


                         


 : 












              
         
                       
   

1-16
AI Education Curriculum for Junior Secondary Schools in Hong Kong



                  
                     
      

         


       
         
         
            
   
       

1 2
So, AI consists of big data, machine
learning and cloud computing. Can I
say that big data is like a knowledge Yes, smart girl!
repository, cloud computing is
similar to our brain power, and
machine learning is like human logic
when we learn?

1-17
Chapter 1: Introduction to Artificial Intelligence (AI)

    Optional WEBSITE

 

    


   



______________________________________________________________________

______________________________________________________________________

 

______________________________________________________________________

______________________________________________________________________

 

______________________________________________________________________

______________________________________________________________________

   

    

                       
             ­
 ­ €  ‚ƒ   €­  
€„ „ ­   ­  ­   
­ ƒ   ­€†   ­ ­   
ƒ  ‡­ˆ €†   ­  ­  ‰ 
­  €

        ˆ­ ˆ  


   €          ­ ­ 
          ƒ              
 ­ Š ­ ­ 
 ‹ŒŽ‘‰      ’ “ 
  € ˆ  ˆ    ­    
      Ž‚€

1-18
AI Education Curriculum for Junior Secondary Schools in Hong Kong

       


       
                     
  ­€    ‚     
    ƒ  „  
 ƒ    ‚            
 ƒ†      ‚ 

‡ˆ ‰   


          ‡        
         
                 
Š   ‹ Œ
           Ž       
 Ž     Ž      Š   
    

‹      


 ‹                  
  ‰   

                 ‘’      
 “   
 Ž       

                  
  ˆ    ˆ 
 Ž    ˆ  



____________________________________________________________________________

____________________________________________________________________________

____________________________________________________________________________

____________________________________________________________________________

1-19
Chapter 1: Introduction to Artificial Intelligence (AI)

1.3 

Learning Objectives


      


       
       

Kelly’s Brother shows Kelly a girl’s picture in his phone.

1 2
Wow, this girl is beautiful! I have downloaded a very funny
Who is she? application on my phone. It can change
my face from a boy’s to a girl’s!
That is…me! Hahaha!
That is really interesting!
But how is it done?

I don’t know …

Are you kidding me?


This is a girl’s picture.

3 3 This application uses AI technologies 4 I heard on the news that a lady was
to recognise facial features, such as scammed by someone with a fake photo
our eyes, nose, and mouth, to change and lost a large amount of money.
people’s appearance.
Yes, I heard that too. AI is a very
powerful technology, but using it
improperly could throw our world
into chaos.

Although this application


is very interesting, I think That’s why ethical
Why, considerations are very
it’s a bit dangerous. Grandma? important when using AI.

1-20
AI Education Curriculum for Junior Secondary Schools in Hong Kong

1.3.1 Ethical considerations of AI 10 BASIC




       


     
                 


     


    
 
    
   

35 INTERMEDIATE
1.3.2 How do AI applications affect our lives in positive or negative ways?

    


         
     
       
                   
 

         ★

­ 

€   ‚

______________________________________________________________________

______________________________________________________________________

______________________________________________________________________

______________________________________________________________________

1-21
Chapter 1: Introduction to Artificial Intelligence (AI)


假新聞AI大師能寫高品質的假新聞,也能識破92%的假新聞

假新聞也不僅由人類產生,機器也可以生產出令人信服的假新聞。隨著時間的推移和技術的發展,這種情
況只會變得越來越普遍,並且可能更有破壞性。真真假假分不清,不如「以火治火」,用AI來檢測AI寫的
假新聞。這個檢測假新聞的AI就是Grover。

你對假新聞司空見慣了嗎?

一張真假難辨的圖片,一個在現實和虛假間游離的故事,這會讓你難以分辨事實的真相。久而久之,在習
慣用質疑的眼光看一切的你看來,真新聞也像假新聞,假新聞也有真事實。

而在今天,假新聞也不僅由人類產生,機器也可以生產出令人信服的假新聞。隨著時間的推移和技術的發
展,這種情況只會變得越來越普遍,並且可能更有破壞性。

真真假假分不清,不如「以火治火」,用AI來檢測AI寫的假新聞。這個檢測假新聞的AI就是Grover。

「我們對Grover的研究表明,檢測虛假訊息的最佳模型就是產生它的最佳模型。」

Grover從Google新聞上5,000個不同媒體撰寫的新聞中進行學習。在這過程中,它接收了120 GB的真實新
聞文章,並用這些文章進行訓練,最終Grover分辨人和AI寫的故事的正確率是92%。在此之前,最好的假
新聞辨識器的正確率是73%。

Grover之所以能如此有效地發現假內容,是因為它也非常善於自己製作內容。畢竟造假大師和鑑假大師
之間只有一線之隔。

華盛頓大學教授和論文的共同作者Yejin Choi在一份新聞稿中就表示辨別假新聞的Grover在假新聞撰寫編
造上也遠比他人得心應手。「在我們的研究中,我們發現Grover的假新聞內容遠比我們人類同胞寫的新
聞更可信,這一事實說明了自然語言生成的發展程度⸺以及為什麼我們需要努力克服這種威脅。」

Grover 研究團隊

這個擅長打假也擅長造假的AI還有能力模仿特定媒體的風格,甚至是特定作家的風格。之前,Grover生成
的一篇假新聞就模仿了《紐約時報》專欄作家Paul Krugman的文章風格。不過研究人員也表示,即使是
最好的人工智慧假新聞,所做的一切依然基於習得的風格和語調,而不是對語言和世界的真正理解。因此
,只要我們有耐心,我們可以從原作者浩如煙海的文章中找到Grover模仿的痕跡。

大方的研究者們將Grover的演示放在網上,每一個人都可以用它寫些天馬行空的假新聞。Fastcompany
的編輯就用它寫了一個《為什麼川普一天要做100個仰臥起坐》的怪誕文章。面對這種專業編輯都有點束
手無策的文章,Grover的表現也蠻出彩:

當你想到仰臥起坐時,第一個想到的形象絕對並不是美國總統。

作為一名三軍統帥,川普的健康狀況幾乎不為人知,雖然他承諾一旦當選總統就會鍛鍊身體。在《名人學
徒》節目中,他對阿諾•史瓦辛格的技術大加嘲諷,沒有什麼能阻止川普不做「川普式的仰臥起坐」。
不過就連川普自己也承認,不管你的工作多麼適合你,要想駕馭自己的身體都是極其困難的。那麼,是什
麼讓三軍統帥走上正軌呢?答案是100個仰臥起坐。

雖然只有短短兩段內容,但是我們也可以看出Grover對川普性格和言論的把握和本人性格非常相似。刨
開令人啼笑皆非的新聞主題,可能真的會有不少人相信這條新聞。

Grover論文的第一作者Zellers表示:「儘管書寫看起來很流暢,但Grover和其他神經語言生成器所寫的
文章包含了機器獨有的語言怪癖,這暴露了它們的機器起源。」這些語言不暢之處就類似於神經文本生成
器的簽名或水印。Grover知道去尋找這些「水印」,這使得它在挑選人工智慧創造的故事時更為高效。

Grover依然需要進化,雖然它已經比同行做得更加出色了,但它依然還有無法辨識的8%假新聞。

隨著人工智慧變得越來越複雜,用AI抵禦AI「製造」可能是我們能夠採取的最佳手段。

1-22
AI Education Curriculum for Junior Secondary Schools in Hong Kong


 


       


   

                   
   

              


   

      

              


­      

      € ‚                  
          
   

          

       ƒ      
 „    €    


   €†      


         
 

          




  €

­     †

‡†    €

                   
            

                  
    ­      €   €        
         

‚€              ­     
    ƒ             
   „            

„ ­         ˆ  ­ƒ          
    

  
               
  ‰       
 † † 

1-23
Chapter 1: Introduction to Artificial Intelligence (AI)

          


        
 

        ­


    
 €  

‚   ƒ     


 „    

       


­      †  
    

         


 

           




‡   €ˆ‰Š       


  

 ‹ Œ     


      ‰ 
 ‰    

‡   ‰ŠŽ‘’      


   

ƒ “       



“                    
          
          
                     
    

1-24
AI Education Curriculum for Junior Secondary Schools in Hong Kong

VIDEO

  ★


   

    


            


_______________________________________________________________________

_______________________________________________________________________

_______________________________________________________________________

_______________________________________________________________________

_______________________________________________________________________



             




   ­€‚ ƒ  „  ­€    


† ‡

ˆ    ‰Š‹ŒŽ ƒ  


  ‘ 

  ‡‚ƒ €       


    ˆ ƒ         Ž    
  € ‡  ƒ  ‚  

’“ ”  ’ˆ


ƒ ‹ŠŠ    ƒ  ”

• ‰Š‹–       


Ž   € 

ˆ  ‚ƒ      


   ‹Š€ „ 
   

1-25
Chapter 1: Introduction to Artificial Intelligence (AI)



AI將改變世界 但「AI偏見」亦會

人工智能偏見,也稱AI偏見(AI bias)已經成為一個熱門話題。在美國,最近爭取種族平等的大規模抗議
活動讓人們注意到治安算法不公平地針對少數族裔,而多年來,已有活動家指出,搜索算法經常將負面
形象或詞語與弱勢社群聯繫起來。在香港版Google中輸入「黑人是」,可以看看世界上最著名的搜尋AI
會給出什麼建議。如今,當人工智能技術將徹底改變我們的社會和經濟時,我們是否應該害怕它也會破
壞我們社會的平等和正義?

可以「學習」,便可有偏見

目前人工智能的絕大部分應用都是基於一類算法,叫做「深度學習」(deep learning) 算法,是機器學習


的分支。深度學習的核心就是識別數據中的特徵,例如照片中對應人臉的特定像素特徵。深度學習算法
之所以大有用處,是因為它可以適應不同數據進行「學習」。算法有不同的「學習」方式,但都涉及在
大規模的訓練數據中自動尋找特徵。譬如,假如我們想讓一個算法在圖像中能夠識別狗,我們就給它展
示許多不同圖像,然後,算法就會嘗試猜測哪個圖像包含狗,哪個沒有;每次失敗,就會調整自己的參
數,直到它能可靠地區分。這些技術的好處是讓我們能夠自動化許多以前只有人類才能完成的任務,譬
如,評估應聘者、評級考試,或者決定哪些罪犯最有可能重新犯罪。結果是,算法對誰被錄用、誰考上
大學或誰被送進監獄的影響愈來愈大。

亞馬遜人工智能招聘引發性別爭議

當亞馬遜開始使用其招聘算法時,算法表現出性別歧視的傾向。為什麼呢?很簡單,因為算法學會了過
去招聘人員的性別偏見。最廣泛討論的偏見類型發生在數據收集期間。

深度學習算法只能從我們提供給它分析的數據中學習,如果訓練數據將偏見的資料輸入演算法,演算法
自然也一定會輸出偏見。電腦科學領域有一句習語指「垃圾進,垃圾出」。「垃圾」可以通過兩種方式
引入到算法中。第一是數據無法很好地反映現實情況:譬如,如果面部識別算法都是用淺色皮膚的臉的
圖片訓練的,那麼它將很難識別其他膚色的臉。第二種情況是,數據正確地反應現實,但現實本身卻是
不公的。這種情況通常發生在使用歷史數據來訓練算法的時候。當亞馬遜公司使用它過去的招聘數據來
訓練其招聘算法時,算法表現出性別歧視的傾向。為什麼呢?很簡單,因為算法學會了過去招聘人員的
性別偏見。可以想像,如果在香港有公司使用歷史數據來訓練招聘算法,它很可能會學會怎麼歧視那些
在底層組別學校上學或住在公屋的求職者。

AI偏見是一面鏡子

我們應該很清楚「種族主義」或「性別歧視」等問題並非來自於深度學習機器本身,而是來自於創造它
的人。我們使用的人工智能雖然強大,但仍然只是一種處理數據的複雜數學工具。就像割草機或吹風機
一樣,深度學習算法是為人製造的,是我們欲望的產物,我們社會結構的結果。AI不會創造新的偏見,而
它會放大已有的偏見。通過教算法如何「學習」,我們教它們像我們一樣行事。通過給它們的目標,提
供的數據或對AI的不同用途,我們教算法重現我們最想隱藏的「特徵」:我們思想和社會的不公平性。
Google會給出帶有種族主義色彩的搜索建議,正是因為人們會搜索種族主義的內容。如果AI公司和研究
員想要學會遏制AI偏見,他們需要審視自己的偏見。古人用「銅、史、人」這三面鏡子來正衣冠、知興替
、明得失,或許在如今這個科技時代,我們也可「以AI為鏡,以祛偏見」。



        


            
                  
   

1-26
AI Education Curriculum for Junior Secondary Schools in Hong Kong

★★

        


                  
    

 

                 
     
       

1-27
Chapter 1: Introduction to Artificial Intelligence (AI)



                  
                        
          

                          
       
                    


1 Brother, you should now under- 2


If the purpose of using AI is to make
stand that AI technologies can
positive contributions to mankind, what
help us, but its improper use can
other ethical considerations should we be
be harmful. Never use technology
paying attention to?
to trick others, do you understand?

Yes, I understand. Let me think about it ...

1-28
AI Education Curriculum for Junior Secondary Schools in Hong Kong

   Optional

  

  

     


   ­ €
  € ‚ƒ„ „ƒ„   
ƒ  „  €ƒ
€   ƒ€„ ƒ  ƒ       ­„    
  

† ‡„ „„ ˆƒ


  ƒ  „ ƒ 
ƒ ­€  ƒ



   ƒ   


ˆƒ€ ƒ€ƒ 
†„€ ƒƒ 

‡€  ‰ ‰  „  „ƒ„ƒƒ  
 €  „ € ƒ      „    Š„
€„



   ˆ ƒ  „         ƒ€  €
„       ƒ      €„  ƒ ƒ   
  †ƒ   „ˆ
ƒ  ƒ€     „
 ƒ„€€„‹‹Œ „ „  
ƒ €ƒ „  
€



  ƒ   †


ƒ   ƒ  €   „ €  € €     
 € Ž
       ­           ‘ ’ “ 
†ŒŽ „   ƒ 
ƒ  €ƒ €
   ƒ    ‚ƒ„     „      
 ” ƒ€   

1-29
Chapter 1: Introduction to Artificial Intelligence (AI)



     


    
       
  

                   
  
  ­€‚ƒ  

„
                 
    ­€‚ƒ  
   

 

†    €   


   ‡ ­ˆ†‰€‚ƒ           
  
       
   Š

   


­ˆ†‰€‚ƒ‹Œ  

______________________________________________________________________

______________________________________________________________________

______________________________________________________________________

______________________________________________________________________

1-30
AI Education Curriculum for Junior Secondary Schools in Hong Kong



 

    _________________________________




   ______________ 


 

_____________________________ 

  _______________________________________________

__________________________________________________________

       __________________       


___________________________________________________

         _____________________ ­€          


‚  _________________

   

  

  
ƒ „ 
    
   

† „  ‡ˆ  ‰Š

 ˆ  
ƒ 
 ‹ 
 Œ 

1-31
Chapter 1: Introduction to Artificial Intelligence (AI)

   

    
 
  
   

       

 ­    


 ­   
 €   
    

‚   

 ƒ „ 
  
 „   
   

† 

‡ˆ‰ Š   


‡‹‰ ­ 
‡‰ ƒ 
‡‰ ƒ  

 ‡ˆ‰Œ‡‹‰  ‡‰
 ‡ˆ‰  ‡‰
 ‡ˆ‰  ‡‰
  „

Ž     

   
  
   
  „

1-32
AI Education Curriculum for Junior Secondary Schools in Hong Kong





___________________________________________________________________________________

                

___________________________________________________________________________________

                 
       

  __________________________________________________________________________

­ ____________________________________________________________________________

€    ____________________________________________________________________

‚    ____________________________________________________________________

ƒ„       

___________________________________________________________________________________

†  ‡     ‡  

___________________________________________________________________________________

1-33
Chapter 1: Introduction to Artificial Intelligence (AI)



 
       
Applied AI
Technologies? Reasons
(✓/✗)
Not applied in normal security checkpoints with X-ray and human operators.
A ✓ or ✗ Applied in the new security checkpoint tool, Syntech ONE, which employs computer vision
(object detection) technologies.

B ✗ RFID tags use electromagnetic waves to transmit ID information to RFID readers.

Calculations by a scientific calculator are executed by logical functions and formulae.


C ✗ It is a human-operated machine used only for math (with the exception of programmable
calculators).
Computer vision technologies are used in self-driving cars to recognise barriers/signals
D ✓
during driving and make corresponding decisions.
Natural language processing technologies are used to understand user inputs and generate
E ✓
responses.
Computer vision techniques are applied to recognise gestures. Speech recognition is also
F ✓
applied to process voice statements.
G ✓ Deep learning is used to recommend songs based on user history.
H ✓ Computer vision techniques are used to detect faces in images.
A 3D printer is like an inkjet printer but for 3D objects. 3D printers can interface with AI, and
I ✗
the two technologies can work together.

      


 Features or functions that are
Features or functions that ARE applications of AI
NOT applications of AI
• Auto-recommendation system: Deep learning • Live chat with salesperson
Taobao • Voice search: Speech recognition • Payment systems
• Image-based waste sorting: Computer vision
• Personalised recommendation: Deep learning • Account login
• Virtual assistant: Natural language processing • Website language interface
Amazon and speech recognition
• Text translation: Natural language processing
• Face recognition camera: Computer vision • Battery charging
• Virtual assistant (e.g. Siri): Speech recognition • Bluetooth connecting
and natural language processing • Volume control
Smart Phones • Automatic image annotation: Computer vision
and natural language processing

• News Feed (personalised recommendation): • Live video streaming


Deep learning • Facebook Messenger
Facebook • Fake news detection: Algorithm
• Lumos (photo identification using image search
system): Computer vision

1-34
AI Education Curriculum for Junior Secondary Schools in Hong Kong

          


Example 1: Description of the failure:
Facial Recognition Failure of facial recognition
Possible reasons causing the failure:
• Dark environment
• Confused by hairstyle, hat or glasses
Example 2: Description of the failure:
Chatbot Failure of chatbot
Possible reasons causing the failure:
• Inputs are too long, complex or ambiguous
• Inputs are out of domain
Example 3: Description of the failure:
World Cup Prediction Failure to predict World Cup results
Possible reasons causing the failure:
• Not enough data on team’s performance
• Many uncertain factors in the competitions

     


      
  Job 1: Reasons:
Driver AI auto-driving can replace the driver.
Job 2: Reasons:
Courier AI Robots or drones can replace couriers.
Job 3: Reasons:
Customer Support Robots with AI voice recognition and face recognition can replace customer services.

  
     
           
        
         
          
 ­  €        
    

     €€  


        ­  €     
     ‚          €  
   
 ­  €        
  ƒ  

     


      „
 Function 1: Function 2:
 User voice records and transcriptions Personalised recommendation
AI technology applied Speech recognition Deep learning
Big data Recorded voices and recognised text (from call centre) User selections and preferences, usage history
Machine learning Neural network transcribing voice into text Neural network analysing recorded user
selections and preferences
Cloud computing Computational power supports the processing of Computational power supports the processing of user
recorded voices and transcribed text history and generating predictions

1-35
Chapter 1: Introduction to Artificial Intelligence (AI)

      


        
Function 2:
Function 1:
Voice input to control the car by voice, useful in
Recognition of barriers and signals
emergency situations
AI technology applied Computer vision Speech recognition
Images and videos of barriers (railings, walls, cars, Recorded voice and recognised text statements
Big data
pedestrians, etc.) and signals (traffic lights)
Neural network analyses images and videos and Neural network transcribes voice statements into
Machine learning
recognises important objects text
Computational power supports the processing Computational power supports the processing of
Cloud computing of images, videos and object recognition recorded voice and text transcriptions to control the
cars.

    


      
        
                
               
     
            
           

 
    
 ­ €  ­ €
        ­     
              
        
     
  €   
  €             

 ‚
 ƒ„ ‚  ‚ 
         ‚   
              
  ƒ   „ †‡         
        „    †ˆ 
              
     ˆ       
            
  ‰   „ƒ †‡       
     Š‹  ƒ         
    Œ     Ž    
            ‘  Œ 
         ­      
          ­  
 Ġ
       ’     
         

1-36
AI Education Curriculum for Junior Secondary Schools in Hong Kong

            


 
• How do you collect enough data for training and testing?
Data Collection
• How do you avoid bias in data collection?
• How do you choose an appropriate model to avoid bias?
Model design
• How do you develop more efficient models?
Computational power • How do you get enough computational power to process data and generate satisfying results?
• How do you prevent AI technologies from misuse?
Ethics
• How do you detect whether the AI technologies are biased?

      


  
    
  
  
  

   


­    ƒ„ ‚ 
 ‹
 
  
    
­  ­
€  €
‚ ‚
ƒ  „

€ ‚ 
 †       
 ‡   ‡ ˆ 
  †‰ † Š
 ‹ ‰ ‡ 
   ‰‡     
 ˆ ‰ˆ ‡  
      
­ ‡     

1-37

You might also like