Artificial intelligence (AI) and machine learning (ML) are currently hot topics in most industries. These buzzwords generally elicit immediate and emotional responses, which tend to fall in opposing camps. One response is characterized by the eternal hope and optimism that AI and ML will allow the automation of mundane or repetitive tasks, freeing humans to focus on more complex work and pushing them to achieve previously aspirational heights—creating a future only imagined in yesteryear’s science fiction. The other? The fear we are ceding our future to robots in a way that will render the human race obsolete as technology evolves.
Typically, these reactions and their resulting worldviews are considered mutually exclusive when it comes to AI and ML’s future impact. However, at Armor we don’t think this is necessarily the case. We believe there’s a third option, one that recognizes the positive impact AI and ML can have on our world and its future while remaining cognizant of the potential risks they pose.
This 2-part blog series will explore 3 topics and questions, including:
- The basics of AI and ML — What are AI and ML? How are they different? What do their definitions tell us about the future? Why are they used interchangeably and how does their usage affect the public’s understanding?
- AI, ML, and cybersecurity — How will AI and ML technology be used in cloud and cybersecurity? Can AI be used by hackers to target companies and individuals, enabling them to steal data? Contrarily, are AI and ML a way for cloud security companies to bolster their defenses?
- AI, ML, and Armor — How is Armor thinking about AI and ML in its pursuit of creating the industry’s leading security-as-a-service (SECaaS) platform?
Today, we’re going to focus on understanding the basics of AI and ML—how they differ and how they work together.
What are AI & ML?
When searching online for AI or ML you’ll find hundreds of thousands of articles using these terms, often interchangeably. These articles feature companies that boast of AI platforms when they’re actually using ML. You will also find a wealth of articles attempting to explain the differences between AI and ML and why those differences matter.
According to Tom M. Mitchell, professor and former chair of the ML department at Carnegie Mellon University, ML is the study of computer algorithms that improve automatically through experience. Simply put, ML is a branch of AI and one of the ways we expect to fully achieve AI. ML relies on working with large datasets, examining, and comparing the data to find common patterns and explore nuances.
There are 3 types of ML that can help clarify what it is, and how it operates
- Supervised learning — training a computer by providing a series of images and data about each image (e.g., X-rays with labels for the symptoms shown). After feeding the machine thousands of images with symptoms, it will be able to take new X-rays and recognize the symptoms.
- Unsupervised learning — occurs when the machine is given a variety of unlabeled images and groups them according to identified patterns without human interference.
- Reinforcement learning — uses observations from interactions with its environment to take actions that would maximize reward or minimize risk. In this case, the reinforcement learning algorithm (aka the agent) continually learns from its environment using iteration. A great example of reinforcement learning is computers surpassing humans and beating them in computer games and chess.
Each type points to a defined scope of what ML is and what it can achieve.
However, AI’s definition is much broader. According to Andrew Moore, dean of the School of Computer Science at Carnegie Mellon University, AI is the science and engineering of making computers behave in ways that, until recently, we thought required human intelligence.
ML is just an AI application based on the idea that machines only require access to data to learn for themselves. Examples of AI are the operating systems in autonomous vehicles and personal assistants such as Google Home, Amazon Echo, etc. These machines perform increasingly complex calculations and tasks in a seemingly human manner. These human-like methods of communicating and performing everyday tasks will become more prevalent in our society and integrate both AI and ML into daily life.
The Resurgence of AI & ML
So, why are AI and ML technologies experiencing a resurgence, and why are these terms used interchangeably? For starters, this is a large attempt by marketers to classify and communicate new and exciting technologies incorporated within their platforms. Since AI is relatively old hat and constantly changes based on new technologies, marketers took the opportunity to revive the term when ML produced results that seemed nearly magical and performed tasks thought to land squarely in the human-only arena.
Second, ML is the latest application of AI. It is not technically wrong to call it AI, but a more accurate description is companies using ML applications to attain actual AI capabilities, not using AI, per se, as a technology. Finally, industry forces have pushed the terms’ adoption as they’re helping to shorten narrowed the cybersecurity skills gap.
Regardless of why and how these terms are used today, it’s important to understand the current state of these technologies, their differences, and understand how ML is helping companies solve complex challenges by providing knowledge in increasingly more human ways.
In our next blog, we’ll turn our attention to how artificial intelligence and machine learning are used within the cybersecurity industry today and how Armor is considering this technology for its leading SECaaS platform. Stay tuned.