A couple of decades ago, self-driving cars, virtual assistants, and smart cities were things found only in science fiction movies. However, today the lines from the reel to the real are blurring. Technology that powers the fabled ‘Iron Man’ suit has made its way into our everyday lives, while hover boards, as seen in the famous ‘Back to the Future’ series, is soon going to be a reality. Technology has ensured that everyday objects become so interconnected and responsive that they achieve contextual understanding and improve the quality of our lives.
Given the enormous value they bring, technologies such as ‘Machine Learning’ and ‘Artificial Intelligence’ are finding a definitive space for their applications in almost all industries – from banking and finance, automobile, defense, aviation, retail, healthcare and more! They are all leaning towards using these new technologies to enable disruption in their industries and rake in better opportunities and profitability. It is because of the huge value that these technologies deliver that this market is expected to be worth USD $2.42 billion in 2017. Given that both ‘Machine Learning’ and ‘Artificial Intelligence’ are all about manipulating enormous volumes of data, many a times, these two technologies are used interchangeably. It is interesting to note that while Artificial Intelligence (AI) and Machine Learning are inter-related , they are definitely not interchangeable!
AI is not a new phenomenon – it has been around for a long time. The term was first coined by John McCarthy in 1965 and uses a broad set of algorithms, technologies and computing methods to make a piece of software more ‘intelligent’ and human-like. AI today is used as an umbrella term that encapsulates everything from actual robotics to robotics process automation and has knowledge engineering at its core. Things like understanding speech, recognizing objects and sounds, planning, learning and problem solving are characteristic of AI. Thus, concepts such as neural networks, natural language processing, robotics, and Machine Learning fall under the purview of AI.
The main aim of Artificial Intelligence is to empower machines with volumes of intelligent data. The aim is to humanize them and give them problem-solving capabilities and consequently make them ‘smart’. In short, AI intends to give machines the capability to achieve ‘human level’ performance for cognitive tasks. For this, the machines have to pass the ‘Turing Test’, for which they need to have automated reasoning capabilities, knowledge representation capabilities and should be able to communicate in a known language to answer questions and arrive at logical and successful conclusions.
Machine Learning also employs large volumes of data to arrive at successful conclusions. The difference from AI, however, rests in ‘how’ the data is used. Machine Learning is a type of Artificial Intelligence that enables software applications to increase their accuracy in predicting outcomes without explicit programming. Machine Learning employs the use of self-learning algorithms that receive input data and use statistical analysis to adjust program actions to predict output values. Simply put, it is a method of data analysis that automates analytical model building so that these models can analyze huge data sets that are more complex and deliver faster and more accurate results, especially on a large scale. Machine Learning helps organizations effectively identify profitable opportunities and helps them avoid unknown risks because of the use of precise data models that uncover connections without human intervention.
Director of Research at The International Institute for Analytics, and a Senior Advisor to Deloitte Analytics, Thomas H. Davenport, states, “Humans can typically create one or two good models a week; Machine Learning can create thousands of models a week.” Using Machine Learning data scientists can apply complex mathematical and statistical calculations to large data sets repeatedly so that they can make high-value predictions.
Nidhi Chappell, head of Machine Learning at Intel says, “AI is basically the intelligence – how we make machines intelligent, while Machine Learning is the implementation of the compute methods that support it.” Once Machine Learning processes data, and makes it more intelligent, AI steps in to use the computational resources at its disposal, for machines to make logical deductions and come up with solutions without the need of human intervention. Using Machine Learning inputs, AI seeks to make machines more intelligent, take real-time actions and make decisions.
Let’s take Google Maps. Google Maps uses AI and Machine Learning to accept an explicit demand of wanting to go from one place to another and aims to satisfy that efficiently, by planning the best route, evaluating traffic, re-routing when a wrong turn is taken etc. AI assistants such as Amazon’s Echo, Apple’s Siri, Google’s NEST, Pandora, Netflix, Cogito etc. use Machine Learning and behavioral science to make these machines intelligent and capable of making human-like decisions. IBM’s supercomputer Watson used Machine Learning to power AI to predict patient outcomes more accurately than physicians.
The success of AI and Machine Learning in real life applications is bringing science-fiction-like technologies to on-the-ground reality. Heavy investments in these technologies to assess how they can be applied to business solutions from Corporates like Google, Facebook, IBM are a testament to the fact that the day might not be far when the real does indeed begin to imitate the reel!