The Difference Between AI and ML
This is part 6 of a 9 part series on Machine Learning. Our goal is to provide you with a thorough understanding of Machine Learning, different ways it can be applied to your business, and how to begin implementations of Machine Learning within your organization through the assistance of Untitled. This series is not by any means limited to only those with a technical pedigree. The objective is to provide a volume of content that will be informative and practical for a wide array of readers. We hope you enjoy and please do not hesitate to reach out with any questions. To start from part 1, please click here.
Terminology and Classification
Throughout this Machine Learning series thus far we have introduced you to a ton of new terminology. This post will be no exception, considering the terms will help us clearly define the difference between Artificial Intelligence and Machine Learning.
First, what is Intelligence? Most definitions of intelligence would be something like “the ability to acquire knowledge and skills and apply it in new environments.” This definition is pretty straightforward, however, it is lacking when we try to use it to define a synthetic or artificial intelligence.
Computers and humans have some similarities with regards to how they gain intelligence. Ultimately, both human and machine intelligence rely on pattern recognition systems to acquire intelligence. But, human intelligence as we know it, gathers intelligence through an array of sensory experience, as well as an additional layer, what we know as consciousness and subconscious perception, which is also informed by sensory experience. This blend creates an acute sense of reasoning and is in part why we have evolved to the top of the animal kingdom hierarchy.
Computers specifically use computational pattern recognition through the ingestion of data to derive meaning from the information being received. In the way that humans decode sensory information to derive meaning, computers use synthetic sensory mechanisms to decode information from data. However, for the time being, computers have not yet fully formulated that secondary layer known as consciousness and blended this layer with their artificial sensory layer.
So what is the point of divergence? Well if I asked you quickly tell me the answer to 14 multiplied by 37, only mathematically minded individuals would be able to rattle of an answer quickly. However, this computation is extremely simple for a computer. Just open your desktop calculator and plug it in. However, if I asked if you could identify if an individual in front of you was upset, you most likely could observe their general body language and tell me the answer. One of these tasks is much more computationally advanced than the other, but a computer would have a very hard time telling you if the human in front of it was upset. We’re barely cracking the surface of a division of special AI called computer vision that programs a computer to be able to recognize the differences between a human and say a stop sign or building. Additionally, the computer is being programmed to tell the difference between a human and a stop sign, but it is far from grasping why it is being asked to tell the difference between the two “objects.”
But why is this? How can a computer easily tell us what 14 multiplied by 37 is, but not be able to derive if a human is upset with the same ease as us? A lot of this comes back to our definition of intelligence. So, we offer a new definition of intelligence. One that can take into account the synthetic, yet fluid nature of Artificial Intelligence. We would define true Artificial Intelligence as all non-biological pattern recognition systems that can extract and transform data into meaningful information while deriving understanding from that information autonomously.
The Origins of AI
Most professionals within the realm of computer science will default to Alan Turing as the creator of the field of Artificial Intelligence. While this may be true from the realm of computer science, the idea of a synthetic intelligence dates back to some of the early existential philosophers, not to take credit away from the “Godfather of AI” .
Philosophers were attempting to devise what truly constitutes consciousness, awareness and the ability to reason with intelligence. When trying to piece apart the components of intelligence, they arrived at the fact that most machines display a level of “special intelligence.” While it can be grasped by a human being, society benefits more by delegating tasks that relate to special intelligence to machines.
For example, the engine in your car could be considered an intelligent machine as it uses sensory information and pattern recognition to drive a complex set of processes under the hood. Most individuals would not be able to tell you how all of the components of a modern car engine work and rightfully so, we don’t have to. All you need to know is how to turn the vehicle on, press the gas and break and you’re good to go. The machine has been endowed with a special intelligence to function in its own right.
However, no one looks at their car and says “there is a synthetic intelligence to marvel at.” So let’s head back to our definition of true Artificial Intelligence: “all non-biological pattern recognition systems, that can extract and transform data into meaningful information and derive understanding from that information autonomously.” Some reading this post may be familiar with the “Turing Test.” It’s hard to write an AI post focused on the composition of intelligence without referring to the legendary concept.
Essentially, the Turing Test is designed to see if a human being can distinguish a computer or synthetic intelligence, from a human or biological intelligence. For our Blade Runner fans out there, this is the test that is issued by blade runners to Replicas, in order to confirm the replica is in fact a synthetic intelligence and not human. The day an Artificial Intelligence can pass the Turing test, is the day that we have arrived at a true AI. This would be the dawn of sentient machines that will quickly supercede the human race. This would be the world’s first true General Artificial Intelligence.
General AI, Special AI, and Machine Learning
Now that you know what classifies as true AI, we’ll introduce you to a spectrum of the current Artificial Intelligence landscape, which will clarify the difference between AI and ML.
On the far left of the spectrum, we have General AI. To classify the AI as General AI, it would need to pass the Turing Test and ultimately be sentient. Additionally, this AI would have some level of synthetic neural plasticity which mimics the human brain’s neural plasticity allowing for the modulation of the AI’s brain to change over time in an advantageous way. Currently, this level of AI would be the most flexible in its intelligence and ability to learn but is the least practical given the current technical requirements and cost of computation.
In the middle of the spectrum would be Special AI. We see this level of AI in some technologies present today, such as self driving cars and Amazon’s Alexa. This AI is computationally advanced, however, it is not sentient (it would not pass the Turing Test). A special AI would get smarter over time and adapt automatically to changes in its respective environments. This level of Artificial Intelligence is moderately flexible and given its presence in some mass market technologies today, relatively practical.
On the far right of the AI spectrum we have Machine Learning. This level of AI offers general pattern recognition, hyper fitted data classification and algorithmic automation. This type of AI is the least flexible on the spectrum. Changes is this AI’s environment means changes in the fundamental composition of this intelligence and re-training to calibrate for the new environment. However, this level of AI is the most practical. It is not computationally expensive to run ML algorithms and they offer a treasure trove of value to the end users. They’re able to automate redundancy while increasing speed and accuracy within the realm of data science.
Sentient Machines Versus Your Marketing Cloud Platform
So you may ask, why is it important to know the difference of Machine Learning and Artificial Intelligence? Although a lot of this post may seem like semantics, understanding the difference will be a great help in the present and future of business. This is especially true for the coming years as we are dealing with a steep adoption curve of Machine Learning and “AI” driven platforms. A lot of companies will attempt to sell you on the great promise of AI, when in reality, their “AI” is comprised of very simple and topical level Machine Learning algorithms. AI is their excuse to charge a premium for their product.
This is the reason why we wrote this post as a part of our Machine Learning series. Artificial Intelligence and Machine Learning have become some of the most common buzzwords used in marketing today and we don’t foresee that changing any time soon. Untitled frequently is pitched by different marketing technology platforms on how we can use these “AI driven marketing platforms” to drive ridiculous value for our partners and clients.
We see right through that and truth be told most companies that have “AI-Driven Stacks” are using very weak Machine Learning techniques. With that said, if you are using a “AI driven marketing platform,” understand what you are buying. We would urge you to pull back the curtains of the tech you are leveraging and get a fundamental understanding of what “super intelligence” is really driving the platform.
In Summary
We hoped you enjoyed this post and will continue on to part 7, our Machine Learning Toolbox, where will take you on an informative journey of the best open source Machine Learning algorithms and libraries.
If you are interested in starting on a Machine Learning project today or would like to learn more about how Untitled can assist your company with data analytics strategies, please reach out to us through the contact form.
Check out part seven of this series.
