Introduction to the Artificial Intelligence (AI) & Machine Learning

Artificial intelligence (AI) spending worldwide was estimated at $37.4 billion for 2019 by analysts at IDC and is expected to grow to $100.1 billion by 2023.  This is an estimate on a broad basis that includes spending on software and services.  Stanford University estimates the U.S. market for AI on this basis at more than $17.6 billion for 2019.  Researchers at Gartner estimated that the global business value derived yearly from AI had already reached $2.6 trillion by 2020, while analysts at PwC estimated that AI’s contribution to the global economy could soar to $13.0 trillion by 2030.
In many ways, artificial intelligence (AI) is a logical extension of recent technology trends.  A combination of ever-more-powerful computer chips, cloud computing, continued miniaturization of devices such as sensors, and the growth of the big data/data analytics sector are major enablers of AI on what is now a cost-effective basis.
Artificial intelligence (AI) and machine learning will create vast changes in nearly all segments of business and industry over the mid-term.  The effect of AI on consumers and households is already in broad evidence, although the people benefitting from such technologies may not be aware of the process or the significance of what’s going on around them.  For example, utilizing machine learning, pioneered the development of advanced software that learns from a shopper’s actions online and then makes product recommendations tailored to the individual.  In its early years, Netflix famously offered a $1 million prize to anyone who could engineer an algorithm that would learn from a subscriber’s movie rental habits in a manner that would increase the accuracy and usefulness of its online recommendation engine by 10%.  The more that Amazon or Netflix can display perfectly curated products for individual shoppers, the happier the consumer and the greater the amount of sales completed.  (Yes, Netflix paid off on this Progress Prize offer, selecting the work of a team of engineers that called themselves “BellKor’s Programmatic Chaos.”)
Search engines like Google and Bing utilize similar technology to serve up billions of dollars’ worth of online ads weekly to carefully-targeted readers of news, entertainment and data online.   These recommendation engines run in the background 24/7; they learn more and more as time goes by and interactions with consumers increase; they benefit from frequent, incremental improvements made by software engineers; and they make the owners of these technologies highly efficient, effective and profitable in their business operations.  
Today, Amazon’s incredibly popular cloud computing subsidiary, Amazon Web Services (AWS), offers the “SageMaker” tool to enable companies of all sizes to quickly build machine learning tools.  AWS also offers easy-to-deploy AI-based tools for speech recognition, image recognition personalization engines, face recognition, forecasting and much more.  Many other firms also offer rapid-deployment AI platforms so that institutions and corporations of all types to easily put AI to work.
Consider the implications of machine learning for critical industrial processes.  For example, airlines around the world spend hundreds of millions of dollars monthly on fuel.  Imagine the benefit, both financially and in terms of reduced carbon emissions, if the air transport sector can reduce fuel usage a mere five percent through the utilization of machine learning—determinizing the most efficient air routes in light of current weather, setting the optimum engine speeds for fuel efficiency and assigning the most efficient flight paths in and out of airports by computer-aided air traffic controllers.  Airlines will thereby reduce both total time in the air and total fuel burnt.  This is but one possibility from tens of thousands of potential applications—virtually all factory, supply chain, and transportation sectors can benefit through such uses of AI. 

Technologies that Have Deep Synergies with Artificial Intelligence
Big Data and Data Mining
Cloud Computing
Digital Assistants (Siri, Alexa, etc.)
Imaging and Facial Recognition
Predictive Analytics
Robotics and Automation
Sensors and Wireless Networks
The Internet of Things (IoT)
Voice Recognition
Source: Plunkett Research, Ltd.

     How AI works:  Simply put, AI and machine learning work by finding patterns in data.  The larger the pool of data, the more observable the patterns and the better the accuracy and outcomes of the machine learning process.  Amazon, for example, not only uses AI broadly in its online services, it is successfully applying it in physical retail stores.  Amazon operates several AI-assisted, brick and mortar convenience store called Amazon Go in major U.S. cities.  Customers may pick up drinks, snacks and prepared meals.  Shoppers scan an app on their smartphones when they enter the store, so that they can be properly identified as individual shoppers.  Cameras throughout the store track shoppers and note which products they have selected while totaling the cost.  The store runs without cashiers, utilizing electronic checkout and payment, while sensors based on AI determine which products were removed from the shelves by which customer, facilitating both checkout and restocking.  There is the potential for a very large rollout of these Amazon Go stores worldwide, with a UK launch possible in the near future.  Technologies refined in the Amazon Go stores may show up in stores at Amazon’s Whole Foods subsidiary, and in its specialty stores that sell books and electronics.  Based on the experience of Amazon and a few other pioneers, AI will have a very significant effect on the way we shop in stores.
One of the more promising advancements is called “deep learning.”  In 2014, Google spent nearly $600 million to acquire UK-based DeepMind, an intensive learning research group.  Deep learning is sometimes referred to in conjunction with phrases such as “machine learning” and “neural networking.”  The main point is that software can be trained by being constantly fed data, queried as to its meaning, and receiving feedback to its responses.  It is essentially training a machine to respond correctly to data of a given nature or to data within a given set of circumstances.

Industry Sectors with Significant Near-Term Benefits from Artificial Intelligence and Machine Learning
(The higher the amount, recency and frequency of data available, the more useful the outcomes from applying AI to such data. Health care is a perfect example, with vast amounts of patient and outcome data captured daily on a global basis.)
Health Care
=         Disease diagnosis and analysis of scans, samples, symptoms and imaging
=         Recommendations for optimum treatment
=         Drug development
=         Enhancement of the “Precision Agriculture” trend, for more effective irrigation, planting and harvesting
=         Prediction of weather and rain
=         Providing traffic flow management
=         Enabling self-driving cars and trucks
=         Optimizing operations for aircraft, truck fleets, railroads and ships
Energy Efficiency and Production/Environmental Controls
=         Developing technologically-advanced “smart cities” and green buildings
=         Improving energy efficiency in air conditioning, lighting and other systems
=         Improving operations and outcomes at all types of energy production operations, from selecting better sites for drilling oil wells to gaining optimum output from windmills
=         Enhancing air and water quality monitoring and control
=         Reducing plant downtime
=         Increasing efficient use of materials and personnel
=         Optimizing actions of robotic equipment
Financial Services
=         Better analyzing risk for insurance underwriting
=         Analyzing optimum investments for specific goals
=         Approving loans and controlling credit risk
Supply Chain
=         Optimizing timing of orders and shipping
=         Reducing inventory wastage and delays
Source: Plunkett Research Ltd.