Lecture

AI Introduction

15 Sep 2022 • Richard Kuo

Introduction to AI: Evolution, Influences, Applications, Future, Hardware, Machine Learning, Deep Learning, Key researchers, Competitions.

AI 簡介

Artificial Intelligence (A.I.)

圖靈測試 (Turing Test)

One-world Turing Test

A one-word Turing Test suggests ‘poop’ is what sets us apart from the machines

Answers from the one-word test show the most popular single answer was “love.” Image by McCoy and Ullman / MIT

AI 的演化

Ref. 三分鐘看懂人工智慧

AI 的影響

機器人來了：人工智能時代的人類生存法則
約翰·普利亞諾（John Pugliano）撰寫了一系列有關人工智能方面的書籍，其中包括《機器人來了：人工智能時代的人類生存法則》。他認為，任何循規蹈矩而且可預測性的工作在未來5-10年都將有可能被人工智能取代，至少在發達國家是這樣。
AI人工智慧取代大量人力，工業5.0時代它如何影響我們的生活？
根據麻省理工學院（MIT）和波士頓大學（Boston University）經濟學家的一項研究，如果AI技術發展速度加快，到2025年，機器人僅在製造業就可以取代200多萬名工人。但隨著AI技術變得越來越智能，人工智能技術更有可能會超越人類，甚至可能會讓專業度很高的職業也被淘汰。
10種職業在20年後恐消失！　AI將取代700萬個工作
哪些工作最可能在未來被人工智慧給全面取代？《websensa》整理了十大未來可能被取代的職業：
1. 電話推銷員（99％）
2. 會計師（98%）
3. 保險和福利經理（96%）
4. 櫃檯接待員（96％）
5. 快遞人員（94%）
6. 零售銷售人員（92%）
7. 校對人員（84％）
8. IT專員（65％）
9. 市場研究分析師（61%）
10. 廣告銷售人員（54%）

AI 的應用

AI 的未來

Autonomous Artificial Intelligence Guide: The future of AI

5 Benefits of AI and autonomous AI

4 industries benefitted from AI

Top 10 Emerging AI And ML Trends 2022

Hyper Automation
Hyper automation is a business-oriented AI trend used by IT organizations to automate processes.
Conversational Artificial Intelligence
Conversational AI combines natural language processing and chatbots or voice assistants to understand human speech and carry out voice tasks.
Latest AI Trends in Medicine
Artificial Intelligence can help medical imaging identify early signs of diseases and spot cancer cells in potential patients.
Latest AI Trends in Education
Personalization is a huge part of AI, where the algorithm is trained to provide the user with personalized content.
Tiny ML
Tiny ML is a fast-growing trend in Machine Learning today. It is utilized in hardware components such as microcontrollers deployed in electric cars, refrigerators, etc.
Quantum ML
Quantum ML is an area of research in quantum computing and machine learning that focuses on converting machine learning algorithms into qubits instead of bits.
AI Conceptual Design
Artificial intelligence is now used to create visual designs by combining language and images from simple text descriptions.
AI in Cybersecurity
Machines identify attack patterns and notify the authorities as potential threats are detected.
Edge AI
Edge AI devices are smart speakers, smart cars, robots etc. Edge AI devices are known for real-time data processing.
AI and ML in the Metaverse
The Metaverse is a virtual space where people can work, play and socialize. One can enter the Metaverse using an AR or VR headset. AI and ML can be incredibly significant in the Metaverse to better user experience.

Artificial General Intelligence (AGI)

An AGI system should have the following abilities:

abstract thinking
background knowledge
common sense
cause and effect
transfer learning

Practical examples of AGI capabilities include the following five

Creativity. An AGI system would theoretically be able to read and comprehend human-generated code and improve it.
Sensory perception. AGI would excel at color recognition, which is a subjective kind of perception. It would also be able to perceive depth and three dimensions in static images.
Fine motor skills. An example of this includes grabbing a set of keys from a pocket, which involves a level of imaginative perception.
Natural language understanding (NLU). Understanding human language is highly context-dependent. AGI systems would possess a level of intuition that would enable NLU.
Navigation. The existing Global Positioning System (GPS) can pinpoint a geographic location. Once fully developed, AGI would be able to project movement through physical spaces better than existing systems.

Meta 人工智慧大神：深度學習並無法實現真正的 AI

他認為現在的人工智慧具備的常識還不如一隻貓，這或許才是解決問題的原點。
楊立昆指出，現在所有的 AI 都面臨一個最基本的難題：「如何測量資訊量」。
楊立昆也提到，產學界更需要重新思考目前 AI 技術面臨的問題，不能只是把海量的數據丟進大型神經網路裡，就希望系統能學會所有東西
身為卷積神經網路（CNN）之父的他坦言，自己已經放棄用生成對抗網路（GAN）技術來預測影片下一幀數的 AI 研究。「那是一次徹底的失敗」他說道。

AI 的硬體

研究：D1 晶片效能達人腦 36%，特斯拉晶片「智力」2023 年超越人類

Tesla AI

Teslas will be ‘more intelligent’ than HUMANS by 2033 as their microchips already have 36% the capacity of the brain, study reveals

Tesla Optimus

Musk discusses details and cost of Optimus robot, working prototype this year

[Homework]: Humanoid Robot with Dexity Hand + AI Vision ?

Tesla is known for making Autonomous Electrical Vehicles (EVs).

Question 1.
Given a Robot price tag at $20K USD, what is the target market that Elon Musk see for Tesla Optimus ?
Question 2.
Given its dexterity, What are the enhanced features (extra arm, replaceable gripper, … ) for a specific application (military, healthcare, miner, …) ?