Table of Contents
1950s - The first AI conference was held at Dartmouth College. It was here that the term "artificial intelligence" was first used.
1960s - Frank Rosenblatt builds the Mark 1 Perceptron computer, based on a neural network that learned through experience.
2000s - DARPA rolls out intelligent personal assistants long before Alexa, Siri, Cortana, and Google Assistant become household names.
2010s - AlphaGO, a deep neural network program from DeepMind, beats Go world champion Lee Sodo in a five-game match. Watch here
What is AI?
Artificial intelligence leverages computers and machines to mimic the problem-solving and decision-making capabilities of the human mind
IBM
In short, AI refers to systems or machines that mimic human intelligence. AI isn't intended to replace humans but is intended to enhance human experience, capabilities, and contributions. AI is able to add value to businesses by providing a more comprehensive understanding to data and automate complex and/or mundane tasks.
AI is a broad field that includes many theories, methods and technologies, as well as the following major sub-fields:
- Machine Learning
- Neural Networks.
- Deep Learning
Sub-Domains of AI
Machine Learning and Deep Learning
ML teaches a machine to make inferences and decisions based on past data while deep learning is a ML technique. To read more about the difference between the two, check out my previous article here:
Deep Learning vs Machine Learning: What's the Difference?
Tina Huynh ・ Mar 27 '22
Neural Networks
Neural networks work on the similar principles as human neural cells. There are three main layers of a neural network: input layer, hidden layer, and output layer.
- Input layer What we want to segregate goes into the input layer
- Hidden layer The hidden layers are responsible for all the mathematical computations or feature extractions on the inputs.
- Output layer The output layer gives us the segregated information desired
Natural Language Processing
NLP is a science of reading, understanding and interpreting a language by a machine. Natural language understanding (NLU) and natural language generation (NLG) are components of NLP but are all dictinct topics. Check out my previous article on NLP vs NLU vs NLG:
Computer Vision
Computer vision algorithms tries to understand an image by breaking down an image and studying the different parts of the object.
Cognitive Computing
Cognitive computer algorithms try to mimic a human brain.
Weak vs Strong
Weak AI
Weak AI is also called Narrow AI or Artificial Narrow Intelligence (ANI), which is AI trained and focused to perform specific tasks.
Strong AI
Strong AI is made up of Artificial General Intelligence (AGI) and Artificial Super Intelligence (ASI). ASI is also known as superintelligence.
Types of AI
Purely Reactive
These machines do not have any memory or data to work with. (ie, chest games)
Limited Memory
These machines have enough memory to make proper decisions but memory is minimal. (ie, suggesting places based on location)
Theory of Mind
Can understand thoughts and emotions. A machine built on this type is yet to be built.
Self-Awareness
These will be the future generation of new technologies! They will be not only intelligent but sentient and conscious.
Different Approaches
Cognitive Modelling Approach
This tried to build an AI model based on human cognition. To distill the essence of the human mind, there are three approaches:
- Introspection: observing our thoughts and creating a model based on them
- Psychological experiments: conducting experiments on humans and observing their behavior
- Brain imaging: using MRI to observe how the brain functions in different scenarios and replicating it through code
The Laws of Thought Approach
Here, there are a large list of logical statements that govern the operation of our minds and can be applied to AI algorithms.
The Rational Agent Approach
According to the laws of thought approach, entities must behave accordingly but there are always some instances where there is no logical "right" thing to do. This is where a rational agent approach tried to make the best possible choice in current circumstances. It is a much more dynamic and adaptable approach.
Conclusion
Where do you think AI will lead us next? Have you done any projects with AI in the past? I would love to hear your experiences and stories in the comments below.
Until next time,
Happy coding!