L1&2 Flashcards
def AI
branch of CS that aims at simulating human intelligence through ML algorithms, so that intelligent machines can reason and make decisions based on data
def Big Data
a set of computational strategies and techniques applied to a large amount of data to mine information from it, including capturing, storing and then analyzing it
difference between AI and BD (1)
BD is the raw material for AI and AI is the learning from datasets to make decisions
Key differences between AI and DS (4)
PTFI (process, techniques, findings, insights)
1. DS is a **comprehensive **process involving pre-processing, analysis, visualization and prediction. AI is implementation of predictive models to forecast future events
2. DS comprises various statistical techniques. AI makes use of computer algorithms
3. DS finds hidden patterns in data. AI imparts autonomy to the data model
4. DS models reflect statistical insights. AI models emulate cognition and human understanding
Strategic importance
of AI (7)
VAEDPEC
1. critical source of business value
2. agility and competitive advantage
3. end-to-end efficiency
4. improved accuracy and decision-making
5. intelligent offerings (products)
6. empowered employees
7. superior customer service
5 V’s of BD
Volume
Veracity
Variety
Value
Velocity
Human Intelligence (7)
- perception of world and oneself
- experience through learning
- problem analyis and resolution
- reasoning, association, judgement and decision making
- linguistic abilities
- discovery, intuition, creativity and innovation
- prediction and insight of development and change
History of AI
1950: alan turing’s “computing machinery and intelligence”
1956: dartmouth conference, birth of AI
1970s: knowledge based systems
1980s: expert systems
1993-2011: rapid development of AI software and hardware
2011-now: modern AI
Narrow AI
- task specific
- majority nowadays
eg. face recognition
General AI
- on par with human capabilities
= strong, deep AI - can think, comprehend, learn and apply intelligence to solve complex problems
Super AI
- surpasses human intelligence
- human sentiments and experiences
- has emotions, beliefs and desires
DIKW pyramid
Data
Information
Knowledge
Wisdom
- intelligence should be at top -> what we’re currently getting to
First-order effects of AI (3)
- process efficiency
- insight generation
- business process transformation
Second-order effects of AI (5)
- operational perf
- financial perf
- market-based perf
- sustainability perf
- unintended and negative consequences
Challenges of AI (7)
BEJADEC
1. complexity and bias of algor
2. privacy, security and ethical regulations
3. poor governance and accountability
4. low quality of data
5. inadequeate expertise, technology and research
6. high cost
7. negative impact on traditional skills, jobs and industries
Expert Systems
- 1980s
- logic and ruse-based approach
- imitates decision-making ability of human expert
- reasons through bodies of knowledge with if-then rules
- used for discrete and highly structured decision making
eg) disgnosing malfunctioning machines, loan decision
ML
- pattern-based approach
- machines can learn and make improvements independently from experience
- computers learn from data
- find patterns and infer own rules
Neural Networks and Deep Learning
- NN composed of artificial neurons or nodes
- deep: hidden layers in multi-layer network
NLP
what is it
eg
- ability to analyse, understand and generate human language and speech
- fills gap between human communication and computer understanding
eg) google translate, spam filtering, customer service
Computer Vision
- gain high-level understanding from digital images or videos
- automate tasks that human visual systems can do
Robotics
- design, construction, operation and use
- control, sensory feedback and information processing by machine
- machines that substitute for humans and replicate human actions
- many are used in places that humans can’t or don’t want to be in
eg) manufacturing, surgical, bomb
