Guavus Blog

All the latest from the world of Guavus

AI vs Machine Learning (part 3 of series)

By: Dr. Roger Brooks, Chief Scientist, Guavus, Inc. and Karina Dahlke, Sr. Product Manager, Guavus, Inc.

There is much confusion in the marketplace regarding what artificial intelligence (AI) really means and how it can help us.  This week we tackle the difference between machine learning and AI and see how both can be applied in business to solve critical problems and foster innovation. If you missed the differentiations between data mining and the categories of machine learning, please check out our past blogs.  Let’s start by defining some terms:

Analytics: 

In general, analytics is the application of mathematics to the analysis of data. In business, we want actionable results. The process generating these results must converge to a set of actionable outcomes. In mathematics, analytical functions possess this convergence property.

Analytics can be broken into various domains, the most popular being data mining, machine learning and now artificial intelligence. With its relatively recent emergence in business, people can use analytics to gain insights to foster improvements or trigger actions that optimize business processes and decisions.

Machine Learning:

Machine learning (ML) is a branch of analytics in which machines continuously improve their ability to recognize patterns as they are trained with more examples, without having to be programmed to handle each example or pattern. Based on the sample data that the machine has been given, it can build models to systematically map and compare new data or situations with past events and patterns and project outcomes.

Sometimes the outcomes from machine learning falsely appear to have been produced by intelligent means. However, in reality, it is just a sophisticated application of pattern recognition.

For example, the first time you heard an app on your smartphone respond with the right answer to your verbal question, you were wowed by what appeared to be its intelligence. In fact, it is doing some very sophisticated pattern recognition or mapping. It is relating your speech patterns to patterns it was trained to recognize. Now if you said something which did not match a pattern it knew about, it would not be able to respond.

Artificial Intelligence:

Artificial intelligence (AI) is a branch of analytics that goes beyond machine learning, providing the system with the ability to reason. In essence, a machine is trying to mimic one aspect of human intelligence. Intelligent systems form a hypothesis from raw, disparate data to develop new information which is not a direct result of the models of data it was provided, or its current knowledge.  Through reasoning, artificial intelligence can create associations between entities or events without ever having seen such maps or patterns before.

Going back to the smartphone example above, if another person heard what you said, but didn’t know exactly what it meant, he or she would extrapolate and hypothesize that such a pattern of words is semantically similar to another pattern of words he or she knows. This is the essence of reasoning and is a key characteristic of intelligence.

AI vs. Machine Learning Illustration:

 Let’s take the same example we studied last week and use it to compare AI and machine learning.

 

In this example, we see that artificial intelligence uses machine learning as a foundational component.

Applying This To Business:

By using ML and AI algorithms to explore relationships between all different types of human and machine-generated events, we can construct models to anticipate what will happen in the future.  The events may be positive, such as predicting where individual customers may want to go after a stadium event and working with nearby restaurants to provide the most appropriate offers.  The events could also be negative, such as predicting when a network, control system or group of sensors will most likely fail in order to take preemptive actions to correct the problem.

We foresee AI differentiating itself when it encounters new events which cannot be directly transformed into known events. For example, what if a product starts exhibiting a new type of failure which is unrelated to any combination of failures in the sample data used to train the system.  This type of failure has not been seen before, yet the machine still needs to interpret the failure and propose a resolution. AI will do this for you, ML will not.

Bringing It All Together:

 At Guavus, we use AI to help our customers dramatically improve their customers’ experiences while reducing the cost of their operations and increasing revenue from their offerings.

 

 

 

ShareTweet about this on TwitterShare on FacebookShare on LinkedInEmail this to someone