AI Flashcards
(38 cards)
what is the definition of AI
The meaning of Artificial Intelligence is the imitation of human intelligence processes by machines. It includes developing systems that can then fulfil tasks which usually require human intelligence – such as speech, decision-making and translation between languages – but without any human intervention.
what is narrow AI
Artificial Narrow Intelligence is AI that focuses on a specific, individual or limited task only.
what is general AI
Artificial General Intelligence (AGI) has not been fully realised, and some developers doubt it will be possible, the machines have to handle a variety of cognitive tasks without much supervision.
what are the different tests for AI
Turing test, Wozniak test/coffee test, Nilsson employment test, Goertzel robot college student test
what is the Turing test?
A test proposed by Alan Turing to assess if a computer can “think” by imitating human responses.
A human interrogator must decide, via questioning, which participant is a computer and which is a human.
If the computer is mistaken for a human, it is said to have passed the test.
what is the Wozniak/ coffee test
-Wozniak’s Coffee Test: inspired by Steve Wozniak.
-A robot must make a decent cup of coffee in an unfamiliar house.
-Involves identifying ingredients, quantities, and performing actions correctly.
-Shows true AI intelligence through real-world problem-solving.
what is the Nilsson employment test
Employment Test: Proposed by AI researcher Nils J. Nilsson.
Suggests replacing the Turing Test with a test based on job performance.
AI must be able to perform human jobs effectively.
Measures Artificial General Intelligence by the percentage of jobs it can fulfil acceptably.
what is the Goertzel robot college student test
AI must enroll in college and earn a degree using the same resources as human students.
Tests AI’s ability to learn, adapt, and perform academically.
Bina48 was the first AI to complete a college class in 2017 at NDNU.
what are the three enablers of the growth of AI
Huags law, Moore’s law, and cloud computing
what is Huags law
Huang’s Law: States GPU performance grows faster than CPU performance.
Contradicts Moore’s Law, which predicts doubling of transistors every 2 years.
GPU performance increases more than double every 2 years.
Driven by hardware, software, and AI synergy.
Graphics processing is essential for advancing AI performance beyond what extra transistors can achieve.
what is Moore’s law
Moore’s Law: Transistor count in chips doubles every ~2 years, while cost halves.
Proposed by Gordon Moore in 1965.
Boosted AI development through increased computing power.
From 1975, doubling occurred roughly every 2 years.
Since 2010, semiconductor improvements have slowed down.
what is cloud computing
The development of fast Internet connectivity along with huge data sets easily distributed and accessed from any point in the connected world enables data-based Artificial Intelligence.
what are the ethical limitations of AI
Bias in Data: AI may not treat everyone fairly if trained on limited or biased data (e.g. car crash tests based only on average-sized men).
Unpredictable Behaviour: AI learns and adapts, so it may not always behave as expected.
Privacy Concerns: People may not want their personal or health data used, even if anonymised.
Incomplete Data: AI decisions may be flawed if the data used is not complete or accurate.
Discrimination Risks: AI may suggest treatments based on race, gender, or age, which could lead to unfair outcomes.
what is machine learning?
Machine Learning (ML) is one of the sub-sets of Artificial Intelligence (AI).
Machine Learning is a specific sub-set of Artificial Intelligence that trains machines how to learn.
Machine Learning is a data analysis method that automates the process of creating analytical models.
what is supervised learning?
You provide a data set that includes pictures of apples.
Then, another data set that tells the models that the pictures are pictures of apples.
Finally, a new set of data that includes pictures solely of apples. At this point, the system can identify the fruit and will remember it.
what is unsupervised learning?
You provide a collection of pictures of different types of fruit.
The model analyses this to try to identify any patterns.
The machine categorises the pictures into different types based on their similar features.
One of these categories would be mainly apples.
what is semi supervised learning
You provide the machine with a data set of pictures and ask it to identify a fruit.
The machine selects a picture of an apple and tells you it is an orange.
You give the system feedback, stating that it is actually an apple. The machine remembers that and selects again, then you give feedback.
This process continues and the machine learns from its experiences and the feedback.
define the term robot
A robot is a machine that a computer can programme that can undertake a complex series of actions automatically.
what is a actuator?
robot component- Small motors directly attached to the machine structure to facilitate movement.
what are end effectors
Equipment that does the actual work and interacts with the environment for the workplace
what are further components of robots?
A Central Processing Unit (CPU) – the robot’s brain.
Sensors — light/sound/temperature.
Power supply — some kind of electrical supply.
Program — providing the logic that stimulates robotic behaviour.
what are the tow methods of programming a robot online
Teaching Tool Programming and teaching by Example
what is teaching tool programming
The robot is programmed using a tablet (Teach Box).
The programmer walks the robot through the process, one step at a time.
what is teaching by example
he operator guides the robot through the process using a manual method. The operator moves the robot using their hands.
The robot records and stores each stage.
