Technology guide 4 - Artificial intelligence Flashcards
What is artifical intelligence?
- Theory and development of information systems able to perform tasks that normally require human intelligence.
- Subfield of computer science
Explain what weak AI is
Weak AI (narrow AI) performs a useful and specific function that once required human intelligence to perform and does so at human levels or better
character recognition, speech recognition, machine vision, robotics, data mining, medical informatics, automated investing, Etc.
What is strong AI?
hypothetical artificial intelligence that matches or exceeds human intelligence.
What is the Turing Test?
computer and human pretend to be women or men, and human has to identify through questions which is human.
Describe the preservation of knowledge for NI and AI
NI: perishable from an organizational POV
AI: Permanent
Describe the duplication and dissemination of knowledge in a computer for NI and AI
NI: difficult, expensive and takes time
AI: easy, fast, inexpensive
Describe the total cost of knowledge for NI and AI
NI: can be erratic, inconstistant, incomplete
AI: consistent and thorough
Describe the documentation of process and knowledge for both NI and AI
NI: difficult and expensive
Ai: easy and inexpensive
HOw does creativity vary between Ai and NI?
NI: can be very high
AI: low, uninspired
Describe the use of senory experiences for NI and AI
NI: direct and rich in possibilities
AI: must be interpereted first; limited
Describe how well NI and AI recognize patterns and relationships
NI: fast, easy to explain
AI: not better than people in MOST cases
Describe the reasoning of NI and AI
NI: makes use of wide context of experiences
AI: good in only narrow, focused, stable domains
What are the reasons for advancement in AI?
- Advancements in chip technology
- Big data
- The internet and cloud computing
- Improved algorithms
What are the stages of AI?
- Recommendation systems
- AI analyzes the data
- Analyze additional data from smart devices and sensors
- Integrate 3 previous steps to enable machines to sense and respond to the world around them
Name 4 AI technologies
Expert systems (ESs)
Machine Learning
Deep Learning
Neural Networks
What are ESSs?
- Expert systems
1. Computer mimics human by applying expertise in a specific domain
2. Either support or replace decision makers
3. Make inferences and arrive at conclusions
What is machine learning?
- Develop algorithms that learn from and make predictions about data
- Build a model from example inputs to make data-driven predictions or decisions
- ML is the ability to accurately perform new, unseen tasks, built on known properties learned from training or historical data that are labelled.
- Process:
1. Have problem
2. Create rule
3. Apply rule
4. Feedback
5. Adjust rule
6. repeat
What is deep learning?
- Subset of machine learning where the computer discovers new patterns without being exposed to labelled historical or training data
- Example applications of deep learning include
- speech recognition,
- image recognition,
- natural language processing,
- drug discovery and toxicology, and
- customer relationship management.
What are neural networks?
- set of virtual neurons that work in parallel to stimulate the way the human brain works
- The neural network assigns numerical values, or weights, to connections between the neurons
- Uses four main components:
1. Inputs
2. Weights
3. Bias or threshold
4. output
Describe current neural networks
- Because of improvements in algorithms and increasingly powerful computer chips and storage, deep learning researchers are able to model many more layers of virtual neurons in neural networks than previously.
- Current neural networks are able to simulate billions of neurons.
What is image processing?
- Each level of neural network manages a different level of abstraction
- To process an image, the first layer is fed with raw image
- That layer notes aspects of the images such as the brightness and colors of individual pixels, and how those properties are distributed across the image.
- The next layer analyzes the first layer’s observations into more abstract categories, such as identifying edges, shadows, and so on.
- The next layer analyzes those edges and shadows, looking for combinations that signify features such as eyes, lips, and ears.
- The final layer combines these observations into a representation of a face.
What is a depp neural network?
- DL automates much of the feature extraction piece of the process
- Eliminates some of the manual human intervention
- Enables the use of large data sets
What are the AI applications
- Computer Vision
- Natural Language Processing
- Robotics
- Speech Recognition
- Intelligent Agents
- Monitoring and Surveillance Agents
- User Agents