The idea of artificial intelligence (AI) has been around for millennia. Stories of AI are scattered throughout history and across the globe: Hephaestus, of Greek mythology, “created” golden robots; Yan Shi (1023–957 BC, Zhou Dynasty) “built” a mechanical man; Leonardo Da Vinci constructed a Robotic Knight. What was once a dream or some crazy figment of one’s imagination, has since become an evolving reality.
Aristotle (384–322 BC) came up with what we could arguably call a simple notion: syllogism. At the core, it relies on logic and reasoning to deduce a conclusion. This is the idea that if A=B and B=C then A=C. The ability to predict outcomes by using what we hold as truths was revolutionary. We still use it today without truly appreciating how groundbreaking this was. Building blocks such as syllogism brought us to the 21st Century realities of AI.
Technological capabilities fostered exponential growth of artificial intelligence into functional realities, scaling beyond theoretical models. Today we use various techniques to achieve AI with multiple levels of intelligence. We have made significant progress since the term artificial intelligence was first coined by John McCarthy (and team) in 1955. Since then our tech has evolved and so has our definition of AI and understanding of its potential. Hold your applause, while we have advanced, we are only at the tip of the iceberg. There are four major categories for AI, possibly more in the future. Today we are only at the tip of Type II.
Let’s take a closer look at the various categories of AI:
AI type I: reactive machines
This type of AI is solely based on logic with no consideration of memory. In a sense it does not remember things from the past that would potentially influence present decisions. An example of Type I AI is IBMs Deep Blue chess playing computer from the 90’s.
AI type II: limited memory
This is where I believe we are today. Machines can look into the past, determine the future and have an understanding of its surroundings. Autonomous Automobiles are examples of Type II AI. These vehicles have the ability to sense its environment and navigate it without human input. Let’s take a deeper look at the current Type II applications:
- Automation (A) – There are many definitions, let’s stick to the most basic. Automation occurs when a machine is processing a particular activity that a human would normally do. An example of automation is the modern home security system that can contact authorities and trigger sprinkler systems in the event of a fire.
- Machine Learning (ML) – This is an application of AI that enables systems to automatically learn and improve their learning over time without explicit programming. These systems focus on the development of computer programs that can access data and use it for themselves. Today’s voice recognition systems such as Siri and Alexa use MI and deep neural networks to imitate human reactions. As they progress these systems will learn to understand the nuances and semantics of our language leading to natural language processing.
- Predictive Analytics (PA) – Depending on its application PA uses a variety of statistical models and machine learning techniques to analyze current and historical facts to make predictions. Broadly, this is used to predict unknown events in the future; however, PA can be applied to any type of hypothetical predictions. A common example of PA is the use of fraud detection mostly seen in financial institutions.
- Natural Language Processing (NLP) – A computer’s ability to process and analyze language. While we have made major advancements in this area, there is still a long way to go. This particular type of AI has been a challenge, and it’s not surprising. Language is extremely complicated. Think back to any language class you may have taken in the past. Yes, English or grammar class counts. This category of AI is extremely difficult to achieve. AI will need to understand both “proper” and “colloquial” phrases & sentences. While challenges exist, companies such as Google, Microsoft, Apple and Amazon are investing heavily in NLP.
AI type III: theory of mind
Theory of mind is a Psychology term that is aggressively being pursued. It refers to the ability to attribute mental states such as beliefs, desires, goals, and intentions to others. It also requires one to understand these states and differentiate from its own state of mind. Computers equipped with a theory of mind would recognize the subject as a conscious agent with a mental world of your own, rather than something purely mechanistic and inanimate. A Fictional example of Theory of mind AI would be the beloved R2D2 in Star Wars.
AI type IV: consciousness
The latest stage of AI is Artificial Consciousness. We have yet to define exactly what this would encompass. Does it include intelligence as well as the emotional components? Does this include the ability to feel (pleasure, pain, happiness, love etc.)? Does it include self-awareness? It maybe all of the above and more.
This seems more science fiction than reality, however the road to reality may be closer than we think. As the idea of AI consciousness evolves from theory to possibility, interest in this area will grow similarly as we see in AI Type II today. There is much debate as to whether AI consciousness poses a threat to humanity and our existence. There is also contemplation as to what threat we may pose to an artificially intelligent agent that achieves consciousness. AI consciousness is indeed the end goal for fictional bots like bicentennial man played by Robin Williams (1999).
Some predict that we are 10-20 years away from Type III and 30-50 years away from Type IV. Hopefully I will experience this development in my lifetime and if it’s just a fraction, that fraction will be revolutionary. Some see advances in AI as a benefit; others say it poses a threat.
Such reluctance and hesitation are driven by the fear of loss or replacement. AI is capable of preforming jobs once held by humans to the point where it has impacted the national revenue stream. Some government entities are considering taxation on bots. Significant advancements in the field and future growth leaves us with exciting anticipation yet brings the potential troubles that come when one leaps into the unknown. Should we be fearful or hopeful in a future where true AI is a real possibility? Only time will tell.