What is Artificial Narrow Intelligence?

by Mansoor Ahmed
Posted: Oct 29, 2020

There is lot of progress in Artificial Narrow Intelligence like smart speakers, self driving cars, AI to do web search and AI application in farming and factory.
The rapid progress in ANI has caused people to conclude that there's a lot of progress in AI, which is true.
But that has caused people to falsely think that there might be a lot of progress in AGI as well which is leading to some irrational fears about evil clever robots coming over to take over humanity anytime now.

There is almost no progress in Artificial General intelligence.
It is the goal to build AI and do anything a human can do.
AGI is an exciting goal for researchers to work on, but it requires many technological break through before we get there.
It may be decades or hundreds of years or even thousands of years away.

What is most important idea in AI?

Machine learning is the most essential idea in Artificial intelligence.
It is a sub set of AI.
Machine learning is a scientific study of algorithms and scientific models that computer system use to perform a specific task without using explicit instructions
Arthur Samuel (1959) has explained the machine learning as " Field of study that gives computers the ability to learn without being explicitly programmed".
Running AI System: A software which automatically returns output B for input A. If we have an AI system running, serving dozens or hundreds of thousands or millions of users, that's usually a machine learning system.

Types of Machine Learning

Supervised Learning

It is the task of learning a function that maps an input to an output based on example input-output pairs.
On one hand,input to output, A to B it seems quite limiting.But when, we find a right application scenario,this can be incredibly value able.
It infers a function from labeled training data consisting of a set of training examples.
In supervised learning, each example is a pair consisting of an input object (typically a vector) and a desired output value (also called the supervisory signal).
A supervised learning algorithm analyzes the training data and produces an inferred function, which can be used for mapping new examples.

Examples A to B mappings

Input (A)? Output (B) Applications

email? Spam Spam filtering

Audio? Text Transcript Speech recognition

English? Chinese Machine translation

image of phone? Defect Visual inspection

Unsupervised Learning

In contrast to supervised learning it is a type of machine learning that looks for previously undetected patterns in a data set with no pre-existing labels and with a minimum of human supervision.
Unsupervised learning, allows for modeling of probabilities densities over inputs.

Reinforcement Learning

It is an area of concerned with how software agents ought to take actions in an environment in order to maximize the notion of cumulative reward.
Reinforcement learning is one of three basic machine learning paradigms.

What enables machine learning to work so well?

Data enables the machine learning to work so well.
The out put of a data science project is a set of insights that can help us to make business decisions.
Data is often unique to our business.
We can acquire data by manual labeling,from observing behaviors of humans,from observing behaviors of machine and downloading from websites.
Don't throw data at on AI team and assume it will be valuable.
Once you have started collecting data, go ahead and start showing it or feeding it to an AI team.
Then the AI team can give feed back to your IT team and what type of data to collect and what type of IT infrastructure to keep on building.
If we have bad data, then the AI will learn inaccurate things.