Using AI and Machine Learning: An Introduction

At Echelon DS, we commonly interface with business leaders who may not be technical but need to understand how Artificial Intelligence (AI) can impact their business. Here we summarize four key articles to provide a high-level introduction to identifying AI opportunities within your existing organization and how to best capitalize on those opportunities. 

Article 1: How to Spot a Machine Learning Opportunity, Even If You Aren’t a Data Scientist

Summary: To understand what a machine learning model does, let’s start with a simple example. Think back to your days of high school algebra, where you learned that the relationship for a straight line is y=mx+b. In those days of high school algebra, you were given m (the slope) and b (the intercept) and used that to draw a straight line between an input x, and an output y. Machine learning in essence reverses the problem: if you have a set of outputs (y’s) and inputs (x’s), what are the m and b that best fit that data? From there, you can extrapolate to the complexities of today’s deep learning algorithms: dealing with multiple x and y variables, non-linear equations, and complex relationships between variables. But at root, all machine learning wants to find the relationship between y and x. When looking for problems in your organization, look for observations you have (your “y’s”) and data that may predict those observations (your “x’s”).

Article 2: What’s the Difference Between Artificial Intelligence, Machine Learning, and Deep Learning?

Summary: Artificial Intelligence (AI), Machine Learning, and Deep Learning are often used interchangeably, and this article gets at the historical sources and meanings of these terms. AI is in fact that oldest and broadest of the terms, reflecting the 1950’s vision of computers eventually replicating human intelligence (think “Rosie the Robot” from the Jetsons, C-3PO, or the Terminator). The article refers to this as “General AI”. Original AI implementations could never reflect this grand vision of human intelligence. They relied on hand-coded “features” of small example data sets, identified by subject matter experts. Machine Learning and Deep Learning are progressively narrower implementations of AI, in which we solve more specific tasks of classification and progressively remove the need for subject matter experts. Subject matter experts are replaced by supplying the computer with ever-increasing example data, and allowing the computer to automatically identify the important features. This progression has been driven by (a) “Big Data” which provides ever-larger example data sets and (b) increased processing power (especially GPU’s that allow for massively parallel processing)

Article 3: The Current State of Machine Intelligence 3.0

Summary: Shivon Zilis’s essential compendium of AI-driven companies is updated annually. While the chart is interesting simply as a listing of relevant machine learning companies, it’s also helpful in examining the business models by which companies are approaching the AI space. Some companies approach the problem by attacking a particular industry (e.g. Healthcare) or Business Function (e.g. HR) and providing machine learning within the context of an application solving a specific business problem. Other companies seek to provide AI tools and technology that form the foundation of AI solutions across industries (see the Enterprise Intelligence and Technology Stack boxes). If you are a company considering using AI within your industry, you can choose a third path: combining the foundational tools and technologies of AI to develop your own solutions that generate a competitive advantage based on your unique insights into current business needs. By utilizing Rho AI, companies can generate their own machine learning technology and gain a leg up on the competition.

Article 4: How Companies Are Already Using AI

Summary: This article talks about finding the first places to apply AI within your current business, and specifically addresses the anxieties that can arise from employees around computers replacing current jobs. In our experience (as well as the author’s), AI rarely replaces current employees, but rather reduces burdens on those employees by removing menial and repetitive tasks, allowing employees to focus on higher-level thinking required to move the business forward. The author suggests looking for opportunities to improve the efficiency of current employees, and to focus first on back-office tasks. Focusing first on removing menial tasks from the plates of employees both improves profitability and increases employee engagement. At Echelon DS, we view data as a fundamental business resource. Adroit data scientists can use just a handful of data points to drive business intelligence that provides a competitive edge. Contact us today to learn more about how we can partner with your business to drive effective data science solutions.