Artificial Intelligence (AI) Simplified
One of the hottest buzzwords in tech right now is artificial intelligence aka "AI". We see AI in popular culture, in movies such as "Star Wars", "Terminator", "Ex Machina", "iRobot" and in the newest HBO series "Westworld" to name just a few.
AI has been around since the 1950s. Through the following decades, the field of AI went through periods of ups and downs as some AI problems proved more difficult to implement than anticipated. This was mainly due to the lack of computing power during those early years. It wasn’t until the late 1990s that AI has seen a resurgence, because researchers focused more on sub-components of AI, and applied the application of AI to solving real-world problems.
But what exactly is AI and what does it do for us in our world today?
AI is technology that learns in two ways. Firstly it learns through what it has been taught and secondly it learns via its own experience. AI never forgets what it learns. This learning combined with experience allows the machine to make decisions, take action and efficiently perform the tasks it has been designed and programmed to complete.
Experts believe that we are still at the early stages of AI development. In a recent "60 Minutes" report, a researcher compared AI to modern flight. He said that if we were to make that comparison, AI of today is as mature as air travel was in the 1930s.
There are many schools of thought about how AI should behave and perform. Some say AI should have the ability to:
- Think like people
- Act like people
- Think rationally
- Act rationally
If true artificial intelligence is the end goal, then rational action is what we should be striving for. The definition of rational action in relation to AI is intelligent machines will work to achieve their goals quickly while using the best possible methods, which means the machines will perform in a manner that maximizes their expected utility.
There are many different fields of AI research. Currently, there is not one AI that can do everything. Different types of AI include:
- Search and optimization - This is experienced when you do web search for a topic.
- Logic and problem solving - Examples include advanced mathematics, physics and defect detection; AI can check millions of "things" quickly for defects compared to what a human can do, without losing focus.
- Perception - Includes object and face recognition and identifying human emotion based upon facial expressions.
- Natural language processing technologies - Includes Google translate, web search and IBM's Watson as seen on the game show Jeopardy. (Spoiler alert - Watson won.)
- Neural networks - The goal of neural networks is to imitate the central nervous system of animals. Its application can be seen in robotics and prosthesis.
- AI portion of Robotics - One example of robotic AI is self driving cars that are currently being tested by Google.
- Game playing - Game playing might be one of the oldest area of AI. Early and current experiments in AI game playing have been demonstrated in games such as chess, checkers, and most recently Go.
- Decision Making - AI decision making has made huge leaps in medical diagnosis, driving directions, web search, email filtering and spam detection, and product recommendations to name a just a few. Medical diagnosis by IBM's Watson is used for cancer treatment. There are thousands of new cancer studies published daily along with new breakthroughs that doctors alone can't keep up with. Watson is able to access and interpret these new research papers quickly and apply them towards designing new treatment strategies for cancer patients.
Conclusion
Although Artificial Intelligence has been around since the 1950's, it is still in its infancy and its fullest potential has yet to be realized. The recent successes in AI have occurred due to dramatically increased computing power and the focus on the various AI sub-disciplines. Following the philosophy of Moore's Law, I expect we will see tremendous advancements in AI in the coming years across all of its disciplines.