Week 6 - Artificial Intelligence in Business Flashcards
What is Artificial Intelligence?
- HUMAN PERFORMANCE - achieving Results - Pursuit of Human-Like intelligence : psychology observations and hypotheses, thought process
- RATIONALITY - doing the Right thing - Rationalist Approach : combination of Maths, Enginerring, Statistics, Control Theory and Economics
Explain - Acting Humanly : Can a Computer Pass for a Human? The Turing Test , 3
- Turing Repose that If a Machine Acts like as Intelligently As a Human Being that it is Intelligent As a Human Being
- a Thought Experiment in which a Human Judge Poses Written Questions
- a Computer Passes If the Human Judge Cannit Tell Whether the Written Responses Come From a Person Or From a Computer
State what 6 DISCIPLES which Compose Most of AI which a COMPUTER NEEDS to PASS A TURING TEST
- Natural Language Processing to Communicate
- Knowledge Representation to Store What it Knows or Hears
- Automated Reasoning to Answer Questions and Draw Conclusions
- Machine Learning to Adapt New Circumstances and Detect Patterns
- Computer vision and Speech Recognition to Perceive the World
- Robotics to Manipulate Objects and Move About
Explain Thinking Humanly : The Cognitive Modeling Approach
- we Must Know How Humans think
can learn about Human Thoughts in 3 Ways : - Introspection : observing Thoughts and Emotions in Action
- Psychological experiments : Observing a Person in Action
- Brain Imaging : Observing a Brain in Action
- Once we Have a Sufficient Precise Theory of the Mind, it Becomes Possible to Express the Theory as a Computer Program
- Cognitive Science Brigs Together Computer Models from AI and Experimental Techniques from psychology to Construct Precise and Testable Theories of the Human Mind
Explain Thinking Rationally : the “Laws of Thought “ Approach
- Aristotle Syllogims - Patterns for Argument Structures that Always Yielded Correct Conclusions when Given Correct Premises
- Codify “right thinking” - Arrange Info in a Logical Order that Offer can Follow
- Law of Thoughts were Supposed to Govern the Operation Of the Mind ; their Study Initiated the Filed called Logic
Explain Thinking Rationally : the “Laws of Thought “ Approach, in terms of Logical Notation, Theory of Probability and Theory of Rational Action
- in Principle, Programs could Solve “Right Thinking” Solvable Problems Described in Logical Notation
- Logic Requires Knowledge of the World - when Knowledge isn’t 100% Certain it’s Not Easy to Represent it in Logical Notation
- Theory of Probability Fills this Gap - Allowing Rigorous Reasoning with Uncertain Information, however, it Doesn’t Generate Intelligent Behaviour
- for that, we Need a Theory of Rational Action - Rational Thought, but Isn’t Enough
Explain Acting Rationally : the Rational Agent Approach
- an Agent is Something that Acts : Operates Autonomously, Perceive their Environment, Create and Pursue Goals, Create and Pursue Goals, Adapt to Change, Persist Over a Long Period of Time
- Rational Agent is One that Acts so as to Achieve the Best outcome, or When there’s Uncertainty, the Best Expected Outcome
- One Way to Act Rationally -> Deduce that a Given Action is Best and then to Act on that Conclusion ; Thinking Rationally. However, this Isn’t Always Correct
- Other hand, there’s Ways of Acting Rationally that Cannot be Said to Involve Inference - recoiling from hot stove is a reflect action that’s usually more successful than a slower action taken a careful deliberation
- can Conclude that Thinking Rationally is 1 Way to Act Rational
Acting Rationally : the Rational Agent Approach - Explain the ADVANTAGES Over Other Approaches
- it’s More General Than the “Law of Thought” (thinking rationally) approach - Correct Inference is just 1 of Several Possible Mechanisms for Achieving Rationality
- is More Amenable to Scientific Development Than Approaches that are Based on Human Behaviour (Acting Humanly) or Human Thought (Thinking Humanly) - Standard or Rationality is Mathematically Well-Defined and Completely General
Foundations of AI - State & Explain 5 Disciplines that Contributed Ideas, Viewpoints and Techniques to AI
Disciplines that Contributed Ideas, Viewpoints and Techniques to AI :
- Philosophy - How does the Mind Arise from a Physical Brain?
- Maths - what can be Computed?
- Economic - How should we do this when the Payoff may be Far in the Future?
- Neuroscience - How do Brains Process Information?
- Psychology - How do Humans Think and Act?
Explain the Difference between AUTOMATION and AUGMENTATION
AUTOMATION : Replacing Humans with AI -> Routinisation (e.g. Warehouse Logistics and Chatbots)
AUGMENTATION : Supporting Humans -> Not Complete Automation (e.g Medical Images diagnoses
Explain why it’s Not Possible to Explain Why an AI Provides Specific Outputs - Relating to Opacity
Reasons why Algorithms can be Opaque :
- Opacity Results from the Complexity of the Algorithmic Code
- is a Result of the Use of Multiple Algorithmic Components which Aren’t All Known or Under the Control of the Designers
- are Instrinisc Characteristics of Learning Algorithms that Redefine Parameters and Relationships with Each Additional Data Point which Make it Hard to Explain and Why a Decision has Been Taken
- Algorithms Have to Tweaked to Understandable Their Results
- these Reasons that Produce Algorithmic Capacity Need to be Acknowleged to be Able to Identify How and When an AI can Add to the Value Organisations Create
Explain the concept of EXPLAINABILITY PARADOX
- the Trade-Off Between the Complexity and Accuracy of AI Models and the Need for Human-Understandable Explanations of Their Decisions and Behaviours
- As AI Models become More Powerful and Sophisticated, they Also Become More Difficult to Interpret and Explain in Human Terms
EXPLIANABILITY PARAOX - Explain How Algorithms’ Structure Problem Solving in Formalised and Pre-defined Rules to be Implemented in Step by Step Computational Operations
- these Rules Aren’t Clear and Easy to Understand : they are opaque
- Algorithmic Opacity has Emerged as a Big Issue to be Addressed to Analyse AI Impact on Value Creation
- if we Don’t Know How AI Works, we Cannot know the Value it Generates
Explain How EXPLAINABILITY PARADOX can CREATE a DILEMMA For Designers and Users of AI Systems
- Creates Dilemma for them, who Need to Balance the Need for Accuracy and Performance With the Need for Transparency and Accountability
- on One Hand, Highly Complex AI Models, e.g. deep neural networks Can Achieve Impressive Levels of Accuracy in Tasks such as Image Recognition or Natural Language Processing
- on Other Hand, they can be Difficult to Understand and Interpret, even for the experts in the field, which Can Raise Concerns About their Reliability, Fairness and Safety
Explain AI VALUE CREATION
- the Mechanisms that Determine the Functioning Of the Algorithm Must be Anlaysed in Nitty Gritty Details to Be Able to Access that Actual Value an AI Generates For an Organisations
- Learning Algorithms that can Adapt Changes In the Environment are By Definition Opaque and Hence their Value is Difficult if not Impossible to Be Determined and Quantified
- AI Reduce the Complexity Of the environment to the Standardised Logic Of the Algorithm. It’s the Environment which Adapts To the Algorithm and Not the Algorithm that Adapts To the Environment
