
Sarb Parhar & Dionne England
What is all the fuss about AI?
It seems like everywhere we turn people are talking about artificial intelligence (AI) and anticipating that it will take over all aspects of how we get things done. New innovations powered by AI seem to pop up daily. There are cars that drive themselves. Cancer detection improvements. The AlphaGo winning game of Go. AI technology has created new perfumes and even paintings. If we want to turn off the lights in our home or play our favorite music, Alexa and Google Home will happily meet our every command.
In this article, we will devel into the basic components of AI and explain why it’s so powerful.
AI is transforming our technology and disrupting our marketplace. To understand how AI works, we first need to understand how humans learn, think and act. AI has the ability to mirror the human learning pattern which is the core of its effectiveness and ensures it will continue to grow in scope and influence.
The three ways that humans learn are:
a) Guided (Aided) learning — where an expert in the subject trains young minds about how to do things. The examples of this are infants being taught to speak at home or how students are taught during their formal education. At the end of each term, students are tested on their knowledge and, if required, they get retraining.
b) Autonomous (Unaided/ Self) learning — an example of this is the new innovation that people come up with based on the associations.
c) Reward-based learning — an example of this is the workplace, where the employee’s performance can be rewarded with a promotion, salary increase, a bonus or all three. Poor performance would not render any of those outcomes and in extreme cases, it could lead to performance management.
What is the present status of AI?
For the first time in the history of humankind, there are mathematical models and algorithms to simulate and replace all three types of learnings – Guided, Autonomous and Reward-Based learning with Supervised, Unsupervised and Reinforcement Learning models. The following table shows the human Learning Styles mapped to real-life examples and corresponding AI models.
Table 1 — Human vs. Machine Learning
Within these three types of models, there are many different algorithms that are available today. Without getting into too many details on these algorithms, the following pictures show how any business problem can be solved with a particlur AI Learning Model.
Picture 1 — Demystifying AI Machine Learning
What is the pre-requisite for AI?
Data, Data, Data…
As you can see in the picture above, Data fuels the AI Engine to solve any and all business problems. To wrap our mind around data it is helpful to compare it with oil, something that we have consumed for the last 100 years. Data is similar to oil with a few of the following caveats:
- Oil is a single-use commodity but Data used by one is still useable by another in a different application. For example, the same Facebook data has been reused by many companies.
- Oil is a finite resource, but Data is not. It can be harnessed and has an unlimited supply.
- Oil in the previous economy was indispensable. In the Intelligence economy, there is a need for data and lots of it.
Why now?
We have gone through various waves of technological developments in the last century, which has brought the technology to the point that the computer can replciate human functions. Humans have few different attributes like long and short term memory, compute capability, power to communicate internally (to assimilate information and come up with new innovations based on the past experiences i.e. data). Interestingly, we have gone through the tehnological progress developing the same capabilities without realizing that we were laying the foundations for replicating human intelligence.
Here are the three waves of technology development as stepping stones along with their implications.
The first wave of technology development was Digitization Age powered by the transistors (and computers).
a) It enabled the disparate information to be saved in the universal language (i.e. binary code). The books, documents, images, music, movies, communication, etc. almost everything we can think of today is digital.
b) In parallel, the invention of transisters coupled with improvements in thin film manufacturing doubled our processing power every 2 years. This observation is called Moore’s Law, which has held true for the last 50 years. As the pace of the growth in processing power was plateauing recently, the Quantum Computing has suddenly given a shot in the arm and catapulted the processing power to the new heights.
All of this resulted in
- Cost of Memory (Data) = 0 (almost)
- Cost of Processing = 0 (almost).
The second wave of technology development brought on by the rise of Communication Age (internet protocol) did the following couple of things.
a) It connected the disparate information being collected from the first wave.
b) It made the offsite processing (i.e. cloud computing) possible and available to masses.
This resulted in
- Cost of Communication = 0 (almost)
- Cost of Search = 0 (almost)
Leveraging the first and the second wave of technology development, we are at the cusp of the third wave of technology development i.e. Intelligence Age, which will connect the processing power with big data using the open Communication protocol leading to wisdom or prediction.
Data + Processing Power + Communication Protocol = Intelligence
With the advent of AI algorithms leading to the Intelligence wave, we have been able to replicate the human learning methods. This will result in
- Cost of Intelligent Servants/ Bots = 0 (almost)
We are already seeing the early signs of these with Apple Siri, Amazon Echo, Google Home, Self-driving vehicles, etc.
Summary and Roadmap
In this article, we looked at the learning styles and attributes of humans vs. AI powered machines. It’s for the first time in history of humankind, we have a technology capable of mimicing different aspect of human intelligence, and in many cases outperforming it.
In the subsequent articles, we will introduce the human decision-making framework, merge the human decision making with AI, and propose new AI algorithms based on human decision making framework.