A wide variety of types of Artificial Intelligence (AI) are revolutionizing business like never before. A PWC survey reported that 86% of participants said that AI will be a “mainstream” technology in their organizations in 2021.
Before you jump into the world of AI, it’s important to understand the different types of AI and their uses. Here’s what you need to know right now.
Intelligence-Based Types of AI
Artificial intelligence is transforming many industries. As technology continues to evolve, AI becomes more accessible to both companies and individuals, but because of the complexity of AI, sometimes we don’t even know where AI exists in our everyday lives.
AI can be divided based on capabilities and functionalities. From a capabilities perspective, there are three types of AI, each with their own characteristics.
The name “narrow AI” represents the capabilities of the machine — narrow and limited. Also known as “weak AI,” these machines operate under many constraints, and are usually designed to focus on a singular task. While narrow AI machines are excellent at what they do, they can’t store data or replicate human intelligence.
Narrow AI is the type we see most in our everyday lives and includes things like facial recognition software, manufacturing robots, Siri or recommended content on streaming platforms.
General AI is sometimes referred to as “strong AI” and this type of AI can replicate some human capabilities. It hinges on the concept that a machine can be provided with general intelligence and then learn over time, eventually gaining the ability to mimic human behaviors. The machines will process and understand the information, then act in the same way a human would.
Unlike the previous type of AI, the challenge with general AI is programming a machine with consciousness. While the technology is evolving, we’re not yet at a point where functions of the human brain can be completely replicated.
Super AI, which has yet to be realized, is where the machines are more capable than humans. Right now, super AI is more of a hypothetical and is the type of AI we often see in movies.
Super AI would require machines to understand and interpret emotions, but that also means there’s the potential for these machines to have emotions of their own, which could be problematic.
Having machines that are better than humans in a multitude of areas, like science, math or medicine, may seem appealing, but we won’t know how this will impact our society until it’s a reality.
Functionality of Types of AI
Under functionalities, we have four types of AI, sometimes referred to as classes I through IV. Let’s quickly look at each one.
Reactive AI is the most basic of the types of AI, as it doesn’t form memories or use previous experience to inform decision making. Reactive machines perceive the world directly, using only the knowledge at hand.
While the machines being used can be effective in specific scenarios, they lack the intuition to interact with the world. Instead, they can be expected to behave the exact same way time after time. Reactive machines can be a good option for testing that an AI system is trustworthy and reliable.
One of the most well-known reactive machines is Deep Blue, the IBM supercomputer that beat international chess grandmaster Garry Kasparov in 1997.
While the overall functionality of limited theory machines is similar to reactive machines, the key differentiator is that they have the ability to observe and store memory. With this type of AI, this is usually achieved by identifying specific data over a period of time. These datasets power the machine and are stored within the memory for future problem solving.
Self-driving cars use limited theory AI and are powered by a combination of what’s stored in the memory and knowledge that has been preloaded into the machine.
Theory of Mind
Theory of mind refers to the psychological term that explains how people, animals and objects around us have their own thoughts and emotions that affect their own behavior. Just like the human mind, theory of mind AI machines are designed to predict and interpret behavior.
While it’s impossible to have a machine replicate the functions of the human brain, theory of mind AI machines can use social interactions to build their learning centers. While other types of AI are currently a reality, theory of mind is still being advanced. It’s currently being used in technology like voice assistants and holds promise for the future.
The final step in the evolution of types of AI is to create machines that are autonomous and can build their own representations. This requires building machines with consciousness, or self awareness.
While self-aware machines are still a ways off in terms of development, the more theory of mind machines evolve, the closer we may get. Self-awareness machines hinge on the ability to recognize, assess and replicate the actions that a human would take, so it’s expected that this technology won’t be reality any time soon — if ever.
Jumping Into the Exciting World of AI
Now that you understand the different types of AI and how they impact the world around you, it’s easy to see the potential for this type of technology, as well as the career opportunities in the field.
If you’re interested in exploring a career in AI, the Udacity School of Artificial Intelligence offers a suite of Nanodegree programs designed to advance your career and position you as a competitive candidate for potential employers.