Here at Opening Bell Ventures, we frequently are being asked: "What is Artificial Intelligence and how do I define it to my team?"
With an overabundance of buzzwords trying to brand anything and everything as "Artificial Intelligence", we thought we would provide our definition and perspective in (relatively) simple terms.
The Intelligence Behind Artificial Intelligence
The expanding capabilities of Artificial Intelligence make it one of the trendiest talking points this year. Especially as Digital Transformations have been getting a huge boost In the Covid-19 era. We therefore think it is important to baseline what fits a true Artificial Intelligence paradigm, and how it differs from what marketing departments want you to believe it is.
Properly understood, the excitement behind AI is warranted — its applications are vast, and its promise to change how industries anticipate and adapt to the needs of customers are well warranted — but it is also surrounded by an atmosphere of vague concepts and jargon that can undermine its true value to one’s business model. Here are the basic principles on how to interpret what AI really is and, perhaps more importantly, what it is not.
Artificial Intelligence and Automation Are Not The Same Thing
“It has become appallingly obvious that our technology has exceeded our humanity.” ~Albert Einstein
It’s easy to see where the mix-up occurs between AI and automation. Both exist to increase efficiency, but accomplish this in different ways with differing levels of human involvement. Automation and software-based algorithms (whether simple or infinitely complex) are rule-following computer programs, streamlining repetitive tasks and processes at scale and speed beyond the capabilities of a human doing the same.
An example would be analyzing the credit risk parameters of a person applying for a loan or a credit card. This involves marrying up numerous data points and calculations to arrive at a “risk score” that denotes the risk a financial organization is willing to take on extending a loan or issuing a credit card to the applicant.
The process of credit analysis or credit risk modeling at its core, involves a combination of data, computational formulas, and conditions to output an answer for which a human is needed to apply discretion: Am I willing to take the risk on this individual or not?
If done in small numbers, the task can be manually performed by individuals using calculators or spreadsheets; however, the volume in which these applications are needed to be responded to yielded a vast amount of databases and algorithms with significant computational power to process millions of these a day. In fact, there is a software algorithm that automates a credit decision every time you swipe your card at the supermarket or buy anything online, running discreetly in the order of millions of times per second.
While powerful, this is still a computational exercise in the realm of automation with potentially Machine Learning capabilities for the algorithm to improve its accuracy based on the vast number of past credit risk scores it has calculated so far.
In fact, we sometimes even allow the software algorithm to render a pass/fail decision on transactions in an automated manner by simply telling an algorithm an acceptable score threshold specific to our organization. Anything that exceeds that risk is either immediately rejected or transferred to a human for further consideration.
In short, we defined the math by which we empower the machine to make a decision on our behalf and set some limits on how far we trust that machine to do so on our behalf.
Over the years, as computational power has increased and the amount of information we can analyze has grown, we have slowly increased the threshold by which we let software make these decisions. We have also managed to repeatedly prove that software can make more accurate decisions on credit-worthiness than their human counterparts.
But we haven’t reached the 100% mark by which we let software make all the decisions in our business without human intervention — as it turns out, mimicking human reasoning accounting for extraneous inputs to the decision, is a hard thing to do. Thus, the excitement surrounding true Artificial Intelligence.
True AI at Work
When we pass that threshold and in a manner of speech “hand over the keys” to the algorithm to make all decisions and act upon them, that’s Artificial Intelligence. This means eliminating human intervention in the most profound manner by which businesses are run, grown, and managed. It is an aspiration but certainly not today’s reality.
In order to do so successfully, the software (or the “machine”) needs to be trusted to make the right decision each and every time and to evolve and learn from making wrong decisions.
“AI will be the best or worst thing ever for humanity.” ~Elon Musk
Intrinsic to our humanity is the ability to evolve from sensory inputs that we experience since we are born. We learn that fire burns fingers, we learn that needles hurt (but that the pain is short-lived), we learn that impact is dangerous, hence we obey traffic rules, and so on.
Most importantly, our thinking matures and in most part is a combination of information, intuition and emotion. The latter two are obviously so subjective that are hard to codify in an 'algorithm'
This intrinsic behavior is still limited when dealing with computer algorithms that are confined to prescribed data sets and preset condition logic. The processing power to deduce and infer is limited in today’s software and hardware, and the concept of intuition is one that is hard to describe, let alone code behaviors for a machine to follow accurately and persistently. Empathy is an integrated component of human decision-making and is something that will continue to differentiate us from "thinking machines".
Intuition (or gut feeling) is a crucial component that separates true intelligence from Artificial Intelligence. This is hard for a data geek like me to write, but no matter how much information we feed an algorithm, it would still lack intuition — it would lack true intelligence. Other data geeks who read this publication would claim that prediction equals intuition but I would strongly disagree.
Yet we do come close, and there are legitimate pursuits today that fit the current thinking of what Artificial Intelligence is or should be — Artificial Intelligence is as much a marketing term now as it is a pursuit in the world of data analytics.
How to Think About AI in 2021
Artificial Intelligence (or AI-like products and solutions) are increasing possibilities for organizations across every industry, from more efficient healthcare diagnosis to beefed-up cybersecurity measures and more complex conversational tasks. Where automation once made executing human-made decisions easier, AI now steps in to take over a bigger piece of the human workforce where relatively simple (to the human brain) and repetitive processes and procedures are concerned.
"AI is probably the most important thing humanity has ever worked on. I think of it as something more profound than electricity or fire." ~Sundar Pichai
AI increases the throughput and accuracy of these tasks in a manner that continuously seeks increased precision and efficiency in a self-taught manner. Like any naturally occurring feedback loop, AI technology today uses feedback in the form of data and successful/unsuccessful results to improve its own set of rules on how it makes decisions. It operates much like human minds do, only faster and at a much larger scale.
The result is a steep rise in AI’s involvement in everyday life. As its underlying algorithms are further refined and the quality of data fed to these programs improves, the real-time applications for AI will continue to fuel efficiency and productivity — if businesses are able to adapt and trust the data over their own sense of intuition, prejudice, emotion, or empathy (for better or worse).
Is AI Here to Stay?
Despite AI’s often incorrect usage in popular media outlets and online listicles, one thing is certain: the workplace of the future will be shaped by Artificial Intelligence’s expanding capabilities, whether it is truly AI or only marketed as such. We as a species are constantly on a quest to improve our existence and productivity through experience-based learning. AI is just another tool in our belt to reach the next level of growth and efficiency. AI has already and will continue to change the rules of the game. Until the next “big thing” comes into focus, AI as we currently perceive it is here to stay.
