By Michael Jarus—Chief Architect, Intradiem
Is there a term that generates more buzz in tech circles today than ‘AI’?
Valued at $60 billion in 2020, the artificial intelligence (AI) market is expected to grow to $1.8 trillion by 2030. Fifty percent of companies use AI in at least one business function. AI Day at Tesla draws millions of livestream viewers each year. Bill Gates calls AI the solution to Africa’s food production challenges.
So, what does it all mean? Does AI represent a revolution? Or is it just the next step on the road of technological progress? I think it’s both. In the headlines, AI sizzles, and in the real world, it solves.
Artificial Intelligence for ‘Thinking Machines’
The term ‘artificial intelligence’ was coined in 1956 by John McCarthy, a researcher who later founded AI labs at MIT and Stanford. McCarthy sought an umbrella term to group a subset of tech-related research efforts into a single field focused on developing ‘thinking machines’ that could simulate human intelligence.
In my role as Chief Architect at a company working to incorporate more AI capabilities into its market offering, I think of AI as a model-based approach to problem-solving, anchored in the scientific method. That means identifying problems, observing facts, formulating hypotheses, testing those hypotheses, analyzing results, and drawing conclusions.
The viability of this method depends in large measure on how much data can be brought to bear on the problem at hand. Historically, organizations have been stymied by their limited ability to mobilize and process enough data to solve most problems. They just didn’t have powerful enough tools.
But when the process works—when we can achieve critical data mass and have the means to process it within a reasonable time frame—the result is a new algorithm. As we all know, algorithms increasingly drive the world we live in. Many consider algorithms to be the embodiment of AI, but I prefer to think of AI as the process we use to create specific algorithms to solve specific problems.
Smart or Simply Speedy?
In any case, the steady and massive expansion of computational power over the past two decades is the thing that has allowed companies to harness previously unimaginable quantities of data. That expansion has been so great that an ordinary laptop or smartphone today could handle the 1969 moon launch with plenty of room to spare. Today we can process more data to test more models and create better solutions, which would have required lifetimes to discover in the good old days.
But is that intelligence, or just speed? Computers can certainly ‘think’ (i.e., process) much faster than humans, but they can’t think at all unless humans first feed them raw material, a hypothesis, to process. This is where a lot of companies go astray, fixing their attention on execution but leaving out purpose, feeding mountains of data into sophisticated computer programs and expecting them to spit out something (anything!) significant. Without a Human in the Loop, results produced by an AI model will only be as focused as the model and the data it’s trained on—and humans are still required to provide the model and train the machine.
The notion that artificial intelligence is really more about execution than sentience also helps deflate some of the fears that AI is a potential threat to humanity. Machine learning (ML)—a major AI subset that involves training software to behave a certain way based on statistical patterns—is behind objects like self-driving cars, IBM’s Watson, and image-recognition software. These programs are trained to behave in a certain manner and cannot go beyond the limitations set by their human trainers.
If computers are destined to destroy humanity, it won’t be because they manage to become sentient; it will be because humans fed them something that enabled them to destroy humanity, like a command to launch nuclear weapons.
Say You Want a Revolution
It’s heady stuff—maybe even revolutionary—and potential applications for this powerful new capability are unlimited. But at the same time, I prefer to think of AI as the latest in a long series of technological refinements that help reduce the time and effort needed to find solutions to our biggest challenges.
AI can solve any problems we can think of, but it can’t think of any problems we should solve. It’s our servant, not our master. It’s helping us advance faster, more thoroughly and more effortlessly than ever before. And as we continue to refine it, we’ll be able to apply it to more and more unsolved problems. The real revolution underpinning this advance, in my opinion, is the massive leap in our ability to process all that data we’ve been generating.
By the way, I don’t believe artificial intelligence will ever become smart enough to replace us. No one wants the human race to be overrun by machines that can do us harm.
So, long live the revolution!
About Michael Jarus, Chief Architect at Intradiem
As Chief Architect, Michael Jarus leads Intradiem’s architecture and design team. With 25 years of experience in architectural engineering, Michael Jarus is responsible for creating the future technical vision for the company and aligning the technical roadmap to the product roadmap. His expertise includes Artificial Intelligence, Business Intelligence, Data Structures, Distributed Systems, Computer Modeling and Simulation. Michael holds a Bachelor of Science in Decision Sciences / Quantitative Business Methods from Georgia State University.