Chapter 13 Flashcards
What is AI?
Related technologies trying to simulate and reproduce human thought behavior
What does computers require which AI can provide?
Knowledge
Perception
Reasoning
Understanding
Cognitive abilities
What must computers be able to do to achieve AI?
-Understand common sense
-Understand facts and manipulate qualitative data
-Deal with exceptions and discontinuity
-Interact with humans in their own language
-Deal with new situations based on past learning
How does AI support decision making?
Computers used to do the following analyses:
-What is? (What happened in the past)
-What if? (What are possible outcomes based on inputs)
Now extending to more intuitive questions
-Why?
-What could we possibly do differently?
-How to fix it?
-When?
What is robotics?
Not intelligent (yet)
Good at repetitive tasks
Usually very specialised (sorting, manufacturing)
What are the advantages of robotics?
Not emotional
Consistent
Work in hazardous environments
No HR issues
Works 24/7
What are soft robots?
-Not conventional hard materials
-Useful in certain fields (Picking fruit, dealing with humans)
What are expert systems
-mimic human expertise
-in specific where humans solved problems in the past
(Identifying illnesses, granting credit)
What are the components of an expert system?
Knowledge acquisition facility
Incorporates new rules and knowledge
Facilitates growth
Knowledge database
Similar to database
Includes rules
Heuristic knowledge
Rules related to the problem
Explanation facility
Explains answer to end-users
What are the uses of expert systems?
- Information management
- Help desks management
- Employee performance evaluation
- Loan analysis
- Warehouse optimization
- Planning and scheduling
- The configuration of manufactured objects
- Financial decision making
- Process monitoring and control
- Supervise the operation of the plant and controller
- Airline scheduling & cargo schedules
When are Expert Systems used?
- Lots of human expertise needed
But difficult for one person to know it all - Knowledge must be representable as rules
- Humans must be able to perform tasks already
- Must be standardized issue with solutions
- Limited in scope (domain)
- Must involve many rules
- Shortage of human experts
What is case-based reasoning?
New problems solved looking at previous solutions to similar problems
-Retrieve past cases
-Reuse solutions
-Revise cases and solutions
-Retain new solution for future use
What are Intelligent Agents?
-Bots / Virtual agents / Intelligent virtual agents
-Appears to be able to reason
-Follows a rules-based process
What are the characteristics of Intelligent agents?
Adaptable
Autonomous
Collaborative
Humanlike interface
Mobile (between platforms)
Reactive
What are some examples of Intelligent agents?
-Shopping agents
Pricecheck.co.za
-Personal agents
Gmail’s suggested responses to emails
Android’s suggested responses to SMS’s
-Data-mining agents
-Siri/Alexa/Google Assistant
What is Fuzzy Logic?
Life isn’t binary
There are ranges and ambiguities
Fuzzy logic work on “degrees of membership”
Appears intelligent
Variety of complexities included in decision or result
What is machine learning?
Knowledge is gained through experience
Computer teaches itself based on experience
Artificial neural networks works on pattern recognition
-Fraud detection
-Opportunity identification
-Speech recognition
-Disease identification
What are genetic algorithms?
-Used to find solutions to optimization and search problems
-Form of artificial intelligence
-Concepts of survival of the fittest and mutation
What is natural language processing?
Communicating with computers in “human” language
-Siri, Alexa
-Analyses text
-Creating documentation or reports (e.g. news paper articles)
-Asking “natural” questions, e.g. “who had the most sales in December” instead of manually doing the filters and reports
-Real time translation or transcription
What do AI and Decision support systems do?
- Decision support systems (DSS’s) benefit from different forms of AI
- Improved decisions
- Identify new opportunities and relationships
- More natural interaction through natural language processing
What is Contextual Computing?
User context is used to provide services and information
-location based
-reminders of events
Set to keep on growing
-personalised experiences
what are the advantages of Robotic Process Automation?
-Cost effective
-Improved accuracy and quality
-Consistency
-Improved analytics
-Increased employee productivity
-Increased customer satisfaction
-Speed
-Reconcile multiple systems
-Versatile
What are the disadvantages of AI and Automation?
Job losses
High initial investment costs
Lack of expertise to implement
What is the framework of ethical issues of AI?
Framework:
Define AI goals
Define complexity of the problem
The more complex, the more likely AI to fail
Determine stability of environment
Instability could lead to AI failure
Define bias. Guard against bias
Determine level of human involvement
What are the Ethical Issues of AI?
- AI bias: (Gender/Race)
- Failure: (can we trust the answers)
- AI will lead to increased wealth inequality
- How will humans react
- Unemployment