The Mechanics of AI

AI or Artificial Intelligence is and will gain the attention of everyone over the next few years as it entrenches itself into everything we know and do. We are living in times of rapid technological advancements, so rapid that keeping up with these technological advances feels like one is running a continuous marathon. AI has captured the imagination of everyone as this science fiction dream turned reality has transformed machines to mimic human intelligence, perform tasks that humans perform, learn from experience and even ‘make’ decisions.

So what’s behind all this AI or Artificial Intelligence?

Artificial Intelligence or AI is “The science and engineering of making intelligent machines, especially intelligent computer programs”. AI is a way of making a computer, a computer-controlled robot, or a software think intelligently, in a similar manner to intelligent humans. The human brain has an average of 70,000 thoughts a day.

Artificial intelligence does not work by magic. It has key components or ‘mechanics’ behind it that support its ‘functioning’. At its core is Machine Learning or simply ML. ML is a branch of AI that focuses on using data and algorithms to enable AI to imitate the way that humans learn, gradually improving its accuracy. Coined by American computer scientist?Arthur Samuel in 1959, the term ‘machine learning’ is defined as a “computer’s?ability to learn without being explicitly programmed”.

A machine learning algorithm is?the method by which the AI system conducts its task, generally predicting output values from given input data. There are four types of machine learning algorithms:?supervised, semi-supervised, unsupervised and reinforcement. Supervised Learning: This involves training an AI on a labelled dataset, where the correct answers are provided. The AI learns to make predictions or decisions based on this training data. With unsupervised learning, the AI is presented with data that has no labels. It then tries to find patterns or groupings within the data on its own. Reinforcement Learning on the other hand requires the AI to use feedback in the form of of rewards or penalties as it performs tasks. Through this approach, the AI ‘learns’ to ‘optimise’. Over time, it learns to optimise its actions to maximise rewards.

?

Machine learning, deep learning, and neural networks are all sub-fields of artificial intelligence (see graphics below). However, neural networks are a sub-field of machine learning, and deep learning is a sub-field of neural networks. Deep learning uses labelled datasets (supervised learning) to inform its algorithm whilst classical/conventional machine learning is more dependent on human intervention to learn; more structured data is required to learn.

Neural networks, or artificial neural networks (ANNs), are more complex as they comprise node layers made up of an input layer, one or more hidden layers, and an output layer. These nodes (artificial neurons), connect to other nodes.

Neural networks (ANNs) rely on the output of each individual node to communicate with other individual nodes through specified threshold values that activate the nodes, sending data to the next layer in the network. A neural network that consists of more than three layers (including input and output) can be considered a deep learning algorithm or a deep neural network. A neural network that only has three layers is just a basic neural network.

AI learns to make predictions or decisions based on the data that is supplied to it or it has access to. We come to witness the rapid growth of data in recent times. Businesses, organisations, government institutions, academic establishments, just about everyone is grappling with the data surge that we coming to refer to as ‘Big Data’. Data is the new currency and the lifeblood of AI. It is like fuel for an internal combustion engine.

There are some key requirements or qualifications that the data must meet for it to be of real benefit to AI models. Data is the representation of facts and has value when used or planned to be used in the future. Data can incorporate business transactions that may include text, numbers, images, videos, graphics or sound. Information, on the other hand, is data in context; without context the data is useless.

As with AI Models, good data is essential. You cannot put sub-standard or ‘bad’ fuel into your tank and expect your engine to run well. Ensuring good data should start at the data input stage as this will enhance the management of good data quality; there is no room for ‘garbage in garbage out’. This may be achieved by using Data Quality Dimensions that are measurable characteristics of data and help in defining data quality requirements (source: Watson Knowledge Catalog).

Data Quality Dimensions can help determine the expected results of data quality assessments. Key elements of Data Quality Dimensions include Accuracy (i.e., data is accurate – within the precision required and refers to the degree of confidence that can be placed in the data), Completeness (requisite information is available), Conformity (data is in the correct format), Consistency (data is not in conflict with other data across systems/processes), Precision (data is provided with sufficient precision), Timeliness (data is known to be sufficiently current, up to date and available) and Uniqueness (no entity exists more than once in the dataset).

Through the approach of good data quality, the AI models can extract more precise and better outputs resulting in a better outcome. The success of AI depends on data. The better the quality of the data, and the more the data, the AI models become more accurate and more humanlike.

As the data becomes more diverse and extensive, the better an AI model can learn and generalise. We have seen that the breakthroughs in AI in recent years have often been accompanied by advancements in data collection methods and improvements in processing capabilities.

Artificial Intelligence (AI) is not magic (sorry David Copperfield) but a combination of data, algorithms, and computing power. We are on a journey with AI to teach machines to learn from examples and make informed decisions. In this journey, we will see the unlocking of the potential that AI has to offer which will continue to have a significant impact on the world we know and will force humanity to adapt, for the better.

As we unlock more of AI’s potential, it will continue to reshape our world, making tasks more efficient, solving complex problems, and pushing the boundaries of what machines can do in the hope of enabling humanity to live in greater harmony and prosperity alongside machines.

?

Sources:

·?????? Forbes

·?????? SAS

·?????? ConsiousMind

·?????? NeuroScience

·?????? Watson Knowledge Catalog

·?????? McKinsey & Company