The mention of artificial intelligence usually sparks conversations around the singularity and robots revolting against humans, but what really is AI and how is it used today?
Applications of AI
Outside of the realm of science fiction, AI is becoming more and more prevalent in our society. The technology can be discreet, such as a mobile phone voice-controlled assistant, or more obviously, such as self-driving cars.
Artificial intelligence programs are widely used over a number of industries. The technology can be used to help doctors to analyse images, control robots in automotive manufacturing and even translate for Amazon’s home assistant, Alexa.
Over the years, the definition of ‘artificial intelligence’ has become much more complex than initially described. This is due to the emergence of machine learning and weak and strong AI systems.
Advancement of AI
Over half a century ago, Minsky and McCarthy - known by their contemporaries as the ‘fathers of the field’ - described AI as any task performed by a program or a machine that requires a human’s level of intelligence to accomplish.
It was agreed that AI systems should display one or more of the following characteristics associated with human-level intelligence: learning, reasoning, problem solving, planning, knowledge representation, perception, motion, manipulation, social intelligence and creativity.
Because of these broad definitional categories, artificial intelligence has come to take many forms and categories began to emerge to separate one type from another.
The simplest forms of AI are designed to perform a single task. These are known as ‘weak’ or ‘narrow’ AI and are usually found in everyday products. This type of AI is defined as an intelligent system that has been taught to complete a specific task without being explicitly programmed on how to do so. This can be seen in Google’s Siri, which can recognize speech or language from the user, but is restricted to searching a user’s phone (or the internet) to answer the user’s queries. This is also seen in search engines, e.g. in suggested products you may like based upon previous purchases or search data.
On the other side of the spectrum are ‘strong’ of ‘general’ AI. This type of AI is able to use all the human behaviours associated with intelligence independent from human commands. This is thought to be the future of AI and the ideal strong AI system will possess the cognitive abilities and understanding of humans while also being able to process all information at a much faster rate. It is believed that these systems would become increasingly better than humans in all areas. The moment at which an AI surpasses human intelligence is known as ‘the singularity’. Thankfully, these sophisticated AI systems are only theoretical and are, for now, confined to the research pages of scientists, engineers and sci-fi novelists.
Interestingly, the most researched type of artificial intelligence is neither of these extremes. Researchers are most interested by the idea of an AI using human reasoning as a guide to make informed decisions, but without needing to replicate it completely. A good example of this is the IBM supercomputer, Watson, which was designed to scan thousands of datasets and make conclusions based on the information it received.