artificial intelligence (abbreviation: AI) noun

the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.

– The New Oxford American Dictionary, Third Edition.

The Oxford Dictionary’s definition of AI sums up the whole entire area of AI perfectly. Although, as it is a summary, it is just surface level.

When learning about Artificial Intelligence, it is important to see three sides to a coin. That is if you see the “heads” face as the good side and the “tails” face as the bad side, that means there is one side left… the edge of the coin. Doing this will allow you to be open-minded about AI’s advantages and negatives and not to be scare-mongered or equally ignorant by concerns.

The Fields of Artificial Intelligence

These are the key definitions to be aware of throughout learning about AI. There are many more definitions. However, we will split those out into separate articles to keep things tidy.

Narrow AI

Narrow Artificial Intelligence is where we are today, this could be crawling a webpage to take note of key information, or playing a game of ‘Go’. Its level of intelligence is small and focused on a set thing. Additionally, it will most likely not understand what it is doing and why it should be doing the said thing.

Machine Learning

Machine Learning (commonly heard as ML) are where computer systems are able to learn without programmed instructions. Instead, they use algorithms and statistical models to analyse and draw inferences from patterns in data.

Machine Learning builds an Artificial Neural Network known as an “AI Model”. This model can then be applied to analyse data it has not been shown before.

To allow machines to make decisions, neural networks are modelled after the human brain where they have neurons connected to each other with associated weights that can increase or decrease the chances of them being activated.

Machine Learning is a sub-field of Artificial Intelligence.

Deep Learning

Deep Learning (DL), the sub-field of Machine Learning, also uses Artificial Neural Networks. However, Deep Learning automates most of any tasks that require human intervention. Thus enabling the use of larger data sets.

DL is special because it can be supervised or unsupervised learning which is whether or not data has been labelled. For example, supervised DL (or ML) will have an image of a chair with metadata for the part of the chair to be labelled as “chair” — this includes all the pixel coordinates. Unsupervised learning is where that image is left unlabelled. The DL model ingests unstructured data and automatically determines the hierarchy of features that distinguish what is different between two types of data (such as a chair or an apple).

Deep Learning requires a large dataset because it has to have enough data to work out differences.

Artificial General Intelligence (AGI)

This is the next leap up from Narrow AI and is where an AI system has a human level of intellectual skill and ability. That is, it can perform any task that a human can do.

It is estimated that AGI will last for a very short period of time, like weeks to months. This is because, as the AI is constantly learning, it will surpass an “Intelligence Explosion” where its learning rate skyrockets, sending it straight to ASI…

Artificial Superintelligence (ASI)

To truly understand the abilities of ASI, we would recommend viewing the reading list below — particularly book two if you are in a hurry). Artificial Superintelligence is where a machine’s intellect surpasses that of humans’ — even humans collectively.

Suggestions on Further Reading

Within the field of Artificial Intelligence, it is vital to be aware of its abilities and ethics. A great reading list for you to begin with on AI is the following list. The first three books are easy to read and have a nice surface-level introduction (although less technical). Then The Age of Spiritual Machines will take it a step further to prepare you for books four through to six.

If at any point throughout reading you feel confused — it is highly likely you skipped something — or did not fully understand something. If you know where that is, go back and read it. This is particularly vital for Superintelligence.

  1. Humans Need Not Apply — Jerry Kaplan
  2. Our Final Invention — James Barrat.
  3. Superminds — Thomas W. Malone.
  4. The Age of Spiritual Machines — Ray Kurzweil.
  5. Superintelligence — Nick Bostrom.
  6. The Alignment Problem — Brian Christian.
  7. The Singularity is Near — Ray Kurzweil.