Types of Artificial Intelligence
We can define Artificial Intelligence (AI) as a subset of computer science. Its main goal is to copy human behavior and intelligence in a machine or a device. That way they will be able to do tasks that generally can be only performed by humans. One of the things AI systems are programmed to do is learning, planning, problem solving, reasoning and decision-making.
These systems are made up of sets of instructions and algorithms. Often, techniques and tools like deep learning and machine learning are the fundamentals of that data processing. Once the machine learning algorithms put computer data into the systems, they then use complex statistical methods to allow the AI systems to learn. Machine Learning enables the AI systems to become better and faster at performing tasks. They are able to do that by not being programmed by humans to do so.
Now that we know this, we can divide the AI systems into three categories.
1. Artificial narrow intelligence (ANI), which has a narrow range of abilities;
2. Artificial general intelligence (AGI), which is on par with human capabilities; or
3. Artificial superintelligence (ASI), which is more capable than a human.
1. Artificial Narrow Intelligence (ANI)
Weak AI (Narrow AI)
This is the first classification of AI systems. Narrow AI systems also known as weak AI, are the first and the only type of AI has been successfully executed so far. It is the type of AI that is created to do one specific task. That is why we can say that it is goal-oriented. It is an intelligent system that completes the task fast and without any errors. For example it can do voice and facial recognition, car-driving, browsing the web.
Although these systems and machines are smart, they come with an obstacle. They can only operate with a narrow set of limits and constrains. That is why this type is often referred as weak AI. This type of narrow AI is not design to copy or imitate the human intelligence. It uses algorithms to simulate how humans behave when doing a specific task. This is all based on a narrow range of parameters and contexts.
Thanks to deep learning and machine learning, AI and especially narrow AI have gotten an abundance of breakthroughs. These types of AI systems now have a whole spectrum of applications. From medicine to smart devices, its use is endless.
Narrow AI works on a simple method. It uses NLP (Natural Language Processing) to do tasks. You can recognize AI in chatbots and like technology. They comprehend human natural speech and language and interact with individuals in a more personal manner.
Based on the type of memory it has, narrow AI can be: reactive or limited. The reactive AI is simple and basic. It doesn’t have any abilities to memorize or store data. This is how they imitate human’s behavior to responding to events without previous experience. The limited AI is a little more advanced. It is packed with storage rooms and has the ability to learn and record historical data. Then it uses that stored data to make and inform decisions.
A large portion of AI systems is limited memory AI. Most machines use an enormous amount of data volume. Once they have collected the data, they use deep learning tools to create a more personalized user experience. This is how VA (Virtual Assistants) works. They read, store and analyze your data and then they personalize novel experiences for you.
Here are some examples of Narrow AI:
• Siri, Alexa, Google Assistant
• Image / facial recognition software
• Disease mapping and prediction tools
• Email spam filters / social media monitoring tools for dangerous content
• Entertainment or marketing content recommendations based on watch/listen/purchase behavior
• Self-driving cars
2. Artificial General Intelligence (AGI)
Strong AI / Deep AI
This is the second classification of AI also know as strong or deep AI. This is a concept of machine learning. It has general intelligence that imitates human intelligence and behavior to solve problems. It has the ability to learn from past experience and then to apply that intelligence to work on problems. What makes AGI incredible is the fact that it is able to learn, think and understand like humans do in given situations.
This type of AI is only hypothetical as many developers and scientists haven’t fully made any machinery conscious by themselves. It is the theoretical level of AI. It moves far beyond than the weak AI.
The ultimate Strong AI has to:
• Reason, use strategy, solve puzzles, and make judgments under uncertainty;
• Represent knowledge, including commonsense knowledge;
• Plan;
• Learn;
• Communicate in natural language;
• And integrate all these skills towards common goals.
3. Artificial Superintelligence (ASI)
This type of AI is the ultimate goal of technology. This is the kind of Artificial Intelligence that not only mimics and understands how human think and behave, but it also becomes aware of itself and exceeds the human mind.
Super-intelligence is now only existent in science-fiction. This concept of AI super-intelligence wants AI to evolve to the point where it is equal to human though process, emotion and experience. But, these systems won’t only understand those emotions but to create their own and evoke their beliefs and needs. This AI would exceed at everything we do and have. It will be able to process things faster and will memorize even better. In short, its capabilities will be far superior to ours. The potential of these beings might seem attractive at first, but it comes with many consequences that we are unaware of.
Although it is easy to program and narrow human behavior and intelligence, creating a machine that is equal to the human consciousness comes with many challenges. Even though we are on a journey to create such machines, the consequences might be greater than the challenge itself. We can only imagine what kind of impact will the new generations of AI bring.