What is Artificial Intelligence?

December 18, 2020

Artificial intelligence refers to the intelligence exhibited by machines. It is a form of computer programming that allows a machine to act or behave as if it were a human being, as in the ability to be aware of itself and its surroundings, to learn, to solve problems, to feel emotions and to communicate with humans (for example, through speech).

The term “artificial intelligence” was coined at a conference in 1956. Since the term was first coined, it has been used to describe a broad range of technologies and capabilities.

AI has come to be used to describe a particular approach or method to the design of computer systems and software, such as the ability to learn or adapt based on experience from interacting with the environment.

AI research is divided into subfields that focus on specific problems or on specific approaches or on the use of a particular tool or towards satisfying particular applications.

The central problems (or goals) of AI research include reasoning, knowledge representation, planning, learning, natural language processing (communication), perception and the ability to move and manipulate objects.

Today, AI is used in the fields that include transportation, finance, health, law, education, retail, defense, government, robotics, and many others.

Three distinct approaches to AI

There are three distinct approaches to AI that have developed since the 1950s:

Approaches that emulate human intelligence. Approaches that are inspired by human intelligence. Approaches that are derived from the concept of artificial intelligence, but that are different from human intelligence.

Approach 1 – Emulation of human intelligence

  • based on the idea that AI research should build on the knowledge of how the human brain functions and how intelligence works.
  • based on the idea that AI research should build on the knowledge of how the human brain functions and how intelligence works.
  • the abstractions of information processing and cognitive functions to build intelligent machines.
  • abstractions of information processing and cognitive functions to build intelligent machines.
  • use symbolic representations for knowledge and intelligence, and focuses on the use of logic and reasoning as the basis for developing solutions to problems.

Approach 2 – Inspired by human intelligence

  • based on the idea that AI research should pursue the development of intelligent machines that are inspired by human intelligence.
  • based on the idea that AI research should pursue the development of intelligent machines that are inspired by human intelligence.
  • the abstractions of information processing and cognitive functions to build intelligent machines.
  • the abstractions of information processing and cognitive functions to build intelligent machines.
  • tends to use symbolic representations for knowledge and intelligence, and focuses on the use of logic and reasoning as the basis for developing solutions to problems.

Approach 3 – Derived from the concept of AI

  • based on the idea that AI research should pursue the development of intelligent machines that are derived from the concept of AI, but that are different from human intelligence.
  • based on the idea that AI research should pursue the development of intelligent machines that are derived from the concept of AI, but that are different from human intelligence.
  • the abstractions of information processing and cognitive functions to build intelligent machines.
  • the abstractions of information processing and cognitive functions to build intelligent machines.
  • tends to use symbolic representations for knowledge and intelligence, and focuses on the use of logic and reasoning as the basis for developing solutions to problems.
what is artificial intelligence

Three common threads in AI are:

Representations: Representations are the mapping from the world into the head. The general idea is that the world is noisy, and complex, and difficult to deal with, so we simplify it by creating a model of the world in our head.

Reasoning: Reasoning is the act of using knowledge to derive new knowledge and make decisions.

Learning: Learning is the process by which we acquire new knowledge.

The goal of AI is to represent the world in a way that facilitates reasoning and learning. This is known as an intelligence. Currently, there are many different representations and reasoning systems in use, but they all share the common goal of creating an intelligence.

Representations

Representations are how the world is mapped into our head. There are two primary ways of creating representations:

Symbolic representations: Symbols are just arbitrary representations for objects or facts. Often, symbolic representations are more flexible and easier to manipulate than other representations.

Sub-symbolic representations: Sub-symbolic representations are not arbitray, but rather, represent the world directly. For example, a sub-symbolic representation of a cat might be a set of pixels that looks like a cat.

Reasoning

Reasoning is how we use knowledge to derive new knowledge and make decisions. There are two primary styles of reasoning:

Logic-based reasoning: Logic-based reasoning is the use of logical rules that can be applied to the representation to derive new facts from existing facts.

Heuristic-based reasoning: Heuristic-based reasoning is the use of common sense and rules of thumb from the representation to make decisions in the world. For example, if you see a cat, and you have the heuristic “Cats like to be pet”, then you may decide that you should pet it.

Learning

Learning is the process by which we acquire new knowledge.

There are two primary styles of learning:

Supervised learning: In supervised learning, the system is provided example inputs (usually in the form of a training set), and a desired output. The system is then trained using a process known as backpropagation. In supervised learning, the desired output may be the actual output of the system, or it may be a label of how the system’s output should be.

Unsupervised learning: In unsupervised learning, the system is not given any example inputs, and no desired output. Instead, the system needs to discover structure in the world. Common techniques for unsupervised learning include clustering and association rule learning.

Symbolic reasoning

Symbolic reasoning is the use of logic to derive new facts from existing facts. It is most useful when the world is very complex, and the representation is very simple. For example, if we want to figure out the probability of a particular outcome of a function, we don’t need to know how the function works. Instead, we just use the rules of probability to figure out the probability of the outcome.

One of the most common logical rules used in symbolic reasoning is Bayes’ Rule:

P(A|B) = P(B|A) * P(A) / P(B)

This equation states that the probability of A given B is equal to the probability of B given A times the probability of A divided by the probability of B . It’s a simple formula, but it can be used to learn all sorts of things from data. For example, the equation can be used to create spam filters, or to learn the probability that a new email is spam.

Bayes’ rule is used heavily in supervised learning. The equation is commonly used to create classifiers that are used in supervised learning. In a supervised learning system, the Bayes’ rule is applied to a set of training data, and the result is used to create a classifier that can be used to classify new data.

Unsupervised learning

Unsupervised learning is the process of learning information from data. Unsupervised learning is useful when the world is complex, and the representation is simple.

One of the most common unsupervised learning techniques is clustering, which is the process of finding groupings in the data. For example, if we look at the following set of data:

3, 5, 7, 9, 11

One thing that we can do is cluster it. The simplest way to cluster is to look at the mean of the data. In this case, the mean is 6.5. We can then use the mean to cluster the data:

3, 5, 7, 9, 11

4, 6, 8, 10, 12

We can then use the clusters to identify patterns in the data. For example, we can see that in this set of data, the odd numbers form one cluster, and the even numbers form another cluster.

Another common unsupervised learning technique is association rule learning. This is the process of looking for relationships between facts. For example, if we have the following facts:

fact1: A, B, C, D, E

fact2: A, B, D, C, E, F

fact3: A, B, C, D, E, F, G

Then we could create an association rule that says:

If A and B are true, then E is also true

This rule can be used to generate a lot of useful information. For example, we could use the rule to find all of the combinations of things that A and B are true.

Advantages and disadvantages

The advantage of symbolic reasoning is that it’s very general. If we know the rules, we can apply them to any situation. For example, if we know the rules of probability we can easily figure out the probability of any event. The disadvantage is that symbolic reasoning is very brittle. If we make a mistake in the rules of reasoning, then we can’t recover the results.

In some situations, symbolic reasoning is an excellent tool. However, in other situations, symbolic reasoning is not very useful. For example, if we have an image of a cat, we can’t just apply the rules of probability to figure out the probability of a cat. Instead, we need to use other tools, like unsupervised learning, to figure out that the image is a cat.

Sub-symbolic reasoning

Sub-symbolic reasoning is the use of representations that are not arbitrary. The idea is that the world is noisy, and that it’s difficult to represent it using simple representations. For example, the world is too complex to represent with simple numbers, so we approximate it with pixels.

One of the most common forms of sub-symbolic reasoning is called a neural network. In a neural network, a representation is created by using a number of layers. Each layer represents a simple abstraction from the layer below it. For example, given the following layer:

Input 1 + Add + Input 2 * Multiply + Input 3 + Add + Input 4 + Add + Output

We could represent the following equation:

a+b*c+d+e

The neural network would create a representation from the equation that could be used to perform the following operation:

(a+b*c) + (d+e)

The advantage of a neural network is that it can represent arbitrary non-linear relationships. For example, if we have the following equations:

a + b*c + d + e = f(a, b, c, d, e)

a + b*c = g(a, b, c)

We can use a neural network to create a representation that can be used to perform the following operation:

f(a, b, c, d, e) + g(a, b, c)

The disadvantage of a neural network is that they are very difficult to create. However, as computers have gotten faster, it’s become possible to create better neural networks.

Other Blogposts

Our blogs purpose is to expand insight and dialogue between us, you as our audience, and other authors.
January 8, 2021
Will AI lead to job loss in the future?

AI has the potential to disrupt the way jobs are done across sectors. It is likely that AI will have an impact on all jobs in the future, but the extent of impact may vary from job to job and industry to industry. Although it is too early to predict exactly what the impact will […]

Read More
January 7, 2021
10 Applications for AI in the Future

AI is already changing the world around us – from the way we interact with machines, to the way we think of them, to even the way we see ourselves. The possibilities of artificial intelligence are endless, and while we’re still just scratching the surface, it’s important to take a step back and see what […]

Read More
December 22, 2020
What Is FinTech?

FinTech is a wide-ranging term that covers all digital innovation in the financial industry. It includes everything from payments, peer-to-peer lending, lending, insurance, investments and savings. FinTech is also a global movement. The digital revolution has already transformed the financial services industry in Europe and North America, and the FinTech revolution is now spreading to […]

Read More
December 18, 2020
What is Artificial Intelligence?

Artificial intelligence refers to the intelligence exhibited by machines. It is a form of computer programming that allows a machine to act or behave as if it were a human being, as in the ability to be aware of itself and its surroundings, to learn, to solve problems, to feel emotions and to communicate with […]

Read More
October 19, 2020
AI in Real Estate

If you are interested in how AI can be used in real estate, you can sign up for my newsletter, and I will send you a complimentary guide to AI and Real Estate. In the meantime, let’s take a look at some of the ways AI is already being used in the real estate industry. […]

Read More
September 12, 2020
How is AI being used in calculating Loans?

Most people are aware that AI is being used to predict risk and make decisions in the field of finance. However, most people are unaware of how this technology is being used. AI is an excellent tool for people who are responsible for making decisions. This is because it makes use of historical data to […]

Read More
August 6, 2020
Can AI predict the Real Estate Market?

Real Estate is one of the oldest and most established markets in the world. It is a market that is considered as one of the most stable and predictable ones in most countries. In fact, it is so predictable that it is considered as the first choice for any investor seeking for stability and safety. […]

Read More
May 7, 2020
Using AI to help sell a House

In the real estate market, artificial intelligence (AI) is being used in many ways to increase the efficiency of realtors. From data analysis to social media outreach, AI is a valuable tool in the hands of realtors. Real estate marketing is a slow process, and if you’re a buyer or a seller, there’s a lot […]

Read More
April 21, 2020
What is the state of Artificial Intelligence?

We are in the very early stages of development of true Artificial Intelligence. AI is a broad term that encompasses many different technologies. One can think of AI as falling into 3 major categories: Weak AI- This is artificial intelligence at its most basic. This is AI that has been programmed to learn from its […]

Read More
menu-circle