Discussing artificial intelligence (AI) in most social settings will conjure up visions of ‘tech taking over the world’ scenarios leading our world toward a grave state of dysphoria. Yet in technical and academic circles, reality sets in and we learn that AI is a set of algorithms that produce specific results without the need for programming specific responses in advance.
AI has been around for well over fifty years. It is only in the past fifteen to twenty years that AI has taken on a greater role in our society. AI can be considered the overarching technology that is used. Directly under AI would be Machine Learning (ML) and directly under ML is Deep Learning:
- Artificial Intelligence – Programmed with the ability to mimic learning and reasoning like humans
- Machine Learning – Algorithms that can learn without being specifically programmed
- Deep Learning – This is a subset of ML. Artificial neural networks are used to learn and adapt using massive amounts of data.
In basic terms, Artificial Intelligence (AI) is an intelligent set of algorithms created by humans, designed to mimic human thinking or human thinking patterns.
AI is fully capable (relatively speaking) of performing tasks intelligently without explicitly being told what to do; and is capable of rationally and humanly (giving the appearance of) thinking and acting on its own.
AI is categorized in three levels:
- ANI – Narrow Artificial Intelligence (this is AI today. All AI is ANI)
- Also known as Weak AI, this is when AI mimics human intelligence and/or behavior but remains isolated to a narrow range of parameters and contexts.
- AGI – General Artificial Intelligence
- To have an AI’s ability to simulate human intelligence and/or behavior to the point it becomes identical to that of a human is known as AGI
ASI – Super Artificial Intelligence
- When an AI not only mimics human intelligence or behavior but in effect surpasses it, at this point it becomes an ASI. This is the ‘level of sapience’ where cinema and books refer to in their tales of benign or hostile takeover of intelligent technology.
Machine Learning (ML) is a field inside the AI structure. Keep in mind that both ML and AI are not mutually exclusive. ML cannot exist without AI. It is ML that has allowed AI to breakout from lab rooms and into realistic applications used today. With the onset of internet 2.0, combined with Big Data analytics, Machine Learning began to make use of AI supplied algorithms. ML needs copious amounts of relevant data to train itself on.
Within the sphere that Machine Learning resides, rests Deep Learning.
Deep Learning approaches Machine Learning problems (including previously used ML legacy methods) with what has become known as a neural network. This term called a neural network is used as Deep Learnings connectivity simulates the interconnectivity of the human brain. The premise is that the connections within the parts of the Deep Learning network are equally as important as each of the parts themselves, as would our brain’s connections between our neurons.
A Deep Learning neural network is made up of layers of mainly non-linear data that is connected as one network to resolve a particular problem. Incoming data will pass into the first layer and will be changed into a simulation of the original data. This transformed data will move through to the following layer and will become another simulation of the simulation of the original data. This process will continue onto the third layer and so on until the data reaches the output layer. These simulations are known as ‘deep representations’ of data. This is why the term ‘deep’ is used in Deep Learning.
An Example of Deep Learning Understanding
In this scenario, the AI would be asked whether a particular picture given to it is either a cat or it is not a cat. The first step would be to input into the AI as many pictures of cats as possible, all labeled as “cat” and other pictures that are labeled “no cat”. No other information will be given to the AI related to any programming that would indicate what a cat is or is not. While we use certain identifiers to label cats (fur, whiskers, tails etc.), AI will be left unto itself to determine its own way to identify what characteristics make up a cat and use those rules to identify new data (pictures) coming in.
For now, AI depends on massive amounts of pertinent data to be used. The more pictures given to the AI as correctly labeled cats, the greater percentage of accuracy the AI will have in recognizing a cat. AI will make the determination based on its own standards (pupil size or shape, eye to ear spatial relationship or another complex set of identifiers).
Deep Learning Benefits to Banking
Fraud Detection: Massive amounts of data can be both collected and analyzed simultaneously. According to Deltec Bank, Bahamas – Deep Learning has the ability to review and immediately flag or stop a transaction during its live event. Banks stand to save millions in detecting fraudulent activity in real-time.
Risk Modeling: Investment banks stand to gain the most as this process helps to regulate a banks financial activities and is crucial when pricing financial investments. The use of Big Data will create copious amounts of data input into risk modeling scenarios resulting in faster problem solving and more importantly, more accurate data to use when making critical or large investment decisions.
Prediction of Customer Lifetime Value
CLV (customer lifetime value) is a predictive analysis of all the value an organization will derive from an entire relationship with any one individual customer. The importance is a major growth measurement tool for companies as it assists in creating and sustaining beneficial relationships with customers that can generate higher returns for the bank, leading to higher growth.
Acquiring and retaining customers will give the bank a 360-degree view of the potential of the customer to use the banks services and efficiently communicate those options to the customer over time.
Google Assistant, Amazon’s Alexa, and Apples Siri are all outputs of Deep Learning algorithms. Photo and face recognition security programs, Tesla’s autonomous driving are other examples of deep learning in the enterprise.
For now, we remain in the AI distinction of ANI, and it will be sometime (and maybe for good reason) before we reach AI’s super intelligence level.
Disclaimer: The author of this text, Robin Trehan, has an Undergraduate degree in economics, Masters in international business and finance and MBA in electronic business. Trehan is Senior VP at Deltec International www.deltecbank.com. The views, thoughts, and opinions expressed in this text are solely the views of the author, and not necessarily reflecting the views of Deltec International Group, its subsidiaries and/or employees.
About Deltec Bank
Headquartered in The Bahamas, Deltec is an independent financial services group that delivers bespoke solutions to meet clients’ unique needs. The Deltec group of companies includes Deltec Bank & Trust Limited, Deltec Fund Services Limited, and Deltec Investment Advisers Limited, Deltec Securities Ltd. and Long Cay Captive Management.