In 2018, a Gartner survey revealed that over a third of the companies surveyed currently use AI technology. AI was listed as the leading disruptive developing technology that would play a major role in reshaping businesses. In the past, companies listed data and analytics as their biggest hurdle.
Artificial intelligence is here and it is growing. Companies and consumers use AI in multiple ways from people simply asking for directions to the best restaurant in their neighbourhood to even systems that detect money laundering networks.
When you think of AI, you may immediately think of Siri or Alexa. But AI has more than just one, flat description. It has categories.
So, what are the 4 types of AI?
The Digital Transformation of AI
AI actually began in literature before science. Stories with a tin man that had no heart (Wizard of Oz) to a human-like robot that impersonated a woman named Maria (Metropolis) implanted the idea of AI in people long before scientists made it a reality.
By the 1950s, though, scientists, mathematicians, and philosophers had the idea of AI deep into their brains and purposed themselves to make fiction into fact.
The first golden age of AI existed form 1950 to 1974. Scientists could transform computers into machines that did more than just complete the command they typed in. They could store information.
This was the first step in the digital transformation of AI.
AI research has gone through significant ups and downs, with an AI winter where little AI research took place. AI winters have occurred where AI technology has stalled out. This is what happened in 1974 when the dark period lasted until 1980 when scientists revived AI tech.
Forty years later, we sit here watching AI boom. Computer geeks and scientists have discovered ways to bring AI to life again and make it charming.
How Artificial Intelligence Works
To qualify as true artificial intelligence, AI systems must do some of the following behaviours:
- Knowledge representation
- Social intelligence
Social intelligence and creativity are the newest features of AI technology. The simplest of AI tech today can learn, and scientists have learned that the more AI learns, the more it grows in knowledge, much like a human mind. It begins to reason more easily and problem-solve more when you give it more information.
What Are the 4 Types of AI?
AI had four distinct divisions. Each category builds in the previous one, creating a more complex type of intelligence.
In the end, the final forms of AI are just a theory. But just as artificial intelligence was a theory in the first half of the twenty-first century, so can the last type of AI become reality.
Reactive artificial intelligence is the most basic type of AI and the first one created. It has no memories from past experiences. Thus it cannot use past experiences to inform its present-day decisions.
Reactive AI, like its name, only reacts to existing situations.
Deep Blue, the chess-playing supercomputer IBM created in the mid-1980s, is the classic example of reactive AI.
This reactive AI could identify a chessboard and its pieces. It understood the pieces’ functions and could predict what moves the opponent would make as well as what moves it should make in response. Deep Blue could beat chess champs that no one else could beat.
For example, between 1996 and 1997 Deep Blue defeated Russian chess grandmaster Garry Kasparov 3 1/2 to 2 1/2 games in a series of matches. Deep Blue was the first computer to defeat a human in chess.
Reactive machines are the simplest of the AI technology out there. They cannot only perform the tasks the programmer has put into them.
Limited Memory Machines
In 2012 the deep learning revolution took place. Previous to the revolution, computers were logic-and-symbol based versions of AI. Deep networks now learn by taking in data and then responding to it.
Deep learning AI means computers see the world like children do, with fresh, innocent eyes. They see everything and get the skills they need to move about their new environments. Their intelligence constantly grows as they draw in more data.
With deep learning tech, scientists developed an algorithm to mimic the brain’s inner workings. This algorithm imitated the way neurons connect in the human brain. Thus deep learning increases the AI machine’s intelligence as it receives training.
Systems like Google Cloud and Amazon’s cloud are great examples of deep learning AI and the AI known as limited memory machines.
Limited memory machines respond to previous knowledge and lessons. They store data and then react to what they’ve already learned.
Autonomous vehicles are a great example of limited memory machines. These self-driving machines respond to environments and the movement of people and other vehicles they can see.
Limited memory allows the AI to have a quick reaction time. As a result, the AI grows in value
Theory of Mind
When machines can make decisions like the human mind, they’ve reached a new level of AI called the theory of mind.
Theory of mind AI machines are a proposed type of artificial intelligence and not a reality yet.
The human brain is complex. It identifies, understands, retains, and then remembers emotions and behaviours. This is how you know how to respond to people.
For a theory of the mind AI type of machine to exist, the AI would need to have this type of complex processing. The machine would have to acknowledge the rapid emotional and behaviour shifts in humans. Then it would have to respond to those fluid behaviours.
Theory of the mind AI type of machines would have to learn more quickly than other forms of AI.
While the theory of the mind AI does not exist fully, some elements of it have been created.
Sophia is a good example of the theory of mind AI. She is a humanoid bot that Hanson Robotics created. She has a human-like face and the ability to use appropriate facial expressions as she identifies and then responds to emotions.
Sophia cannot have a full-blown conversation yet. However, her ability to understand and then respond to emotions quickly is a step in the right direction
Artificial Super Intelligence and Self Awareness
The top of Maslow’s hierarchy of needs is self-actualization or self-awareness. This is the peak of AI as well. When scientists figure out a way to create AI with self-awareness, they’ve replicated human-like intelligence.
Artificial Super Intelligence (ASI) is a theory of AI that science has yet to fully achieve. ASI machines will be aware of the mental state of others around them and of themselves. This type of AI is on par with human intelligence in that it mimics emotions, desires, or needs.
Scientists do not currently possess the hardware or algorithms for this type of AI.
Self-aware AI does not just react quickly and appropriately to emotions around them. They have emotions and desires themselves, separate from others around them.
Self-aware AI takes the theory of the mind types of AI up a notch. Theory of the mind AI focuses on recognizing and mirroring human actions. Self-aware AI means the machine can and will guide its own thoughts and reactions
Risks and Benefits of Artificial Intelligence
With all innovation comes risks and benefits. We can see the benefits already from artificial intelligence.
It increases efficiency in everyday life. For example, autonomous cars have the ability to lower accident rates. AI tech like Alexa allow adult children to check on their elderly parents.
But with increasing maturity of AI come increased risks with AI. Security and ethical risks remain as well as the financial risk of companies investing in an AI that might not improve their company overall. Many people worry their jobs are at risks as well with the increase in AI.
However, considering the fact that experts are now saying emotional intelligence is the future of work, AI has a ways to go to replace all workers.
IQ, EQ, and AI
When you answer the “what is AI” question, you can point to multiple examples in our society today. AI is all around us, from the voice-command radio in our homes to the cars we drive to the tech that can identify faces on security footage.
Now you can answer the question, “What are the 4 types of AI?” You also understand that what some people fear has not yet come into reality.