Building AI using ML - Part 1¶

Artificial Intelligence — A Clear and Simple Introduction: Part 1¶

Five part introduction series on “Artificial Intelligence”, an attempt to explain the fundamentals in simple language without much technical jargon.

Before understanding the modern techniques used to build Artificial Intelligence, in the first part of this series, lets us deep dive into “What is Artificial Intelligence?”.

Humans have always sought to employ machines to execute various tasks. In the early days, the primary goal was to delegate the majority of repetitive tasks to machines, thereby allowing individuals to concentrate on tasks that require intelligence.

Examples of this outsourcing can be observed in practices such as printing, which eliminated the need for manual copying, and the development of media recorders and players, which spared individuals from the need to repeatedly perform activities like singing, dancing, and acting.

In the meanwhile, there were studies to understand if the intelligence tasks can also be outsourced.

What is Intelligence tasks?¶

Humans possess the capacity to carry out tasks through cognitive functions, which are acquired either inherently or through experience. The human brain employs one or more of these cognitive functions to engage in tasks that necessitate the processing (comprehension) and retention (memory) of the surrounding environment. The primary cognitive functions encompass perception, learning, reasoning, decision-making, and problem-solving.

Studies aimed at investigating the potential outsourcing of comprehension and memory capabilities led to the development of computing devices, which encompass calculators and computers. At the heart of these devices lie processors and storage components.

The information assimilated from the environment was conceptualized as ‘data’. Although the exact mechanism of data storage in the brain remains unclear, the advent of the binary system and transistors facilitated their representation in binary format.

Perception¶

Perception is the ability to recognize, organize, and interpret information obtained from the environment through sensory input. This capacity is not exclusive to humans; a majority of living creatures on Earth possess sensory mechanisms for environmental awareness. These senses encompass Sight, Sound, Smell, Taste, and Touch. Absolutely, the brain is the entity responsible for processing and interpreting sensory input, enabling the perception of the environment.

From a machine’s perspective, the primary goal in Artificial Intelligence is to bestow machines with the capability for perception. This encompasses both equipping them with sensors and enabling them to interpret the data gathered by these sensors.

Artificial sensors must possess the capacity to convert environmental stimuli into a binary format. While not all challenges in this regard have been fully addressed, significant progress has been made in the realms of Sight, Sound, and Touch. Cameras, microphones, and touch sensors have demonstrated the ability to recognize and organize sensory input data in binary form. Storage technology has also advanced considerably for these senses, employing various formats and compression techniques.

However, challenges persist in the realms of Smell and Taste; as of today, research in these areas remains ongoing. The ultimate challenge for Artificial Intelligence lies in its ability to interpret this information and derive meaningful observations from it.

Olfactory sensors and gustatory sensors are employed to detect and identify smells and tastes, respectively.

Actuation¶

Just as living creatures have the capacity for sensory input through sensors, they also possess the ability to output, such as producing sound or engaging with their environment using organs like hands and legs. This action, known as actuation, is the counterpoint to perception and is executed by devices called ‘actuators’.

Examples of natural actuators include the tongue, which produces sound, and the hands and legs, which facilitate movement. These actions are initiated by commands received from the brain, underscoring that actuation is also a cognitive function of the brain.

Various actuators have been developed to enable machines to replicate similar functions. For instance, speakers allow machines to reproduce sound, while motors and levers serve to mimic the actions of hands and legs. These technological advancements have significantly expanded the capabilities of machines in terms of both perception and actuation.

Reasoning and Decision Making¶

In addition to perception and actuation, the brain triggers numerous cognitive functions, including reasoning and decision-making. These functions do not necessarily rely on external organs. Reasoning involves the capacity to draw conclusions based on available information and established rules. Decision-making, on the other hand, is the process of selecting the most suitable action or response based on learning and reasoning. It can be either rational or irrational, and may stem from explicit or tacit knowledge and beliefs.

All of these functions necessitate some form of learning. While some may argue that these functions can be implemented based on predefined ‘rules’, it’s important to note that these rules are also acquired from past experiences, either on an individual level or collectively as a species.

Natural Languages¶

Beyond the fundamental five senses, there exist additional noteworthy abilities. Many living creatures have developed ways to communicate with each other using these senses, forming an additional layer above basic perception and actuation.

Humans employ various forms of expression, such as natural languages, music, and drawing, to convey messages. These creative abilities are diverse and vary among different species. For instance, birds, animals, and even certain insects like bees have evolved unique methods of communication. Birds sing, while bees use intricate dances to convey messages to their counterparts.

These creative abilities also necessitate unique cognitive functions. Natural languages encompass skills like reading, writing, speaking, and listening. The human brain has honed this ability to a high degree, utilizing it extensively for expressive and comprehensive communication.

Intelligence Vs Artificial Intelligence¶

All of the mentioned tasks are carried out by the cognitive functions of the brain, collectively referred to as Intelligence.

Intelligence encompasses the comprehension and memory capabilities of the brain, which encompass various cognitive and mental functions. A machine, regardless of its sensors, actuators, and peripherals, can be classified as Artificial Intelligence only when it is capable of replicating and executing these functions to a degree comparable to living creatures, especially humans.

What it takes a machine to become an Artificial Intelligence? Let us continue in the next part. (hint: It’s not just machine learning)

Read the second part here.