The value and utility of artificial intelligence is a hot topic today. In fact, a recent report by MIT Sloan Management Review asserts that almost 85 percent of executives believe AI “will allow their companies to obtain or sustain a competitive advantage.”
But there’s also a lot of confusion about terminology, the kinds of applications businesses can make use of, and where AI is ultimately heading. In my experience, it’s not uncommon to hear business professionals use “AI” and other terms — such as “machine learning” and “deep learning” — interchangeably when there are, in fact, distinct differences between them.
AI is the next era of computing, and it’s important for the business world to understand just how to unlock its potential. In a recent Facebook Live event, I talked with Amy Webb, professor of strategic foresight at NYU Stern School of Business and founder of the Future Today Institute, about how AI is not a silver bullet and how real value comes from a tailored combination of tools designed to address a specific business problem.
The stakes are high for companies looking to invest in AI, and incomplete understanding can lead to unsuccessful implementation and lack of ROI. So just what do all these terms actually mean?
The academic pursuit of AI came about after World War II, and the technology was originally intended to replicate the human mind. This included efforts to mimic how we perform general reasoning, value-assessment, and internal inquiry. As time passed, researchers realized that our brains are much more complicated than had been previously supposed. And while human beings are good at what they do, there are things we simply don’t have the capacity for. For example, a doctor can’t read and understand the millions of articles published by PubMed as quickly as an AI system can. With AI technologies, the goal is to enhance human capabilities, thought, and reach, not to replicate or replace them. Just as construction workers use bulldozers and backhoes instead of simple tools, AI allows us to dig through tremendous amounts of data, uncover patterns, and make better decisions based on those insights.
The ultimate goal for AI is what I like to refer to as augmented intelligence, which will help inspire humans to leverage insights for better ideas and alternative perspectives but will also help us see through our cognitive biases to points of view that might not have otherwise occurred to us.
Machine learning (ML) is an algorithmic technique that has been around for decades. On a basic level, ML can predict trends or recognize patterns in data if researchers expose the model to previous examples of those patterns. For example, if you have a series of numbers that you can associate with something significant, and if that pattern has previously repeated itself, you can teach ML algorithms to recognize that pattern and then use it to predict whether that same significant outcome is going to happen again.
For businesses, the implications of ML technology are significant. With ML, enterprises can evaluate buyer behavior to inform future expectations, enabling leaders to adjust strategy if and where necessary. Businesses can deploy ML for a variety of processes. Retail stores can use machine learning in forecasting systems that must consider both past and current (up to the hour) market trends. And financial services can use ML-powered product recommendation systems that must leverage current interest rates, trends, and market movements.
Deep learning (DL) is a more recent innovation under the broader umbrella of machine learning algorithms. DL is an algorithmic technique modeled after the neural structures in human brains. The term is sometimes referred to as “neural networks” because researchers were inspired by the synapses and neurons in the human brain and the mechanism by which neurons fire and cause synapses to collect and propagate the energy of that neuron.
DL uses artificial neural networks to dramatically increase the number of dimensions of data it can work with – including unstructured images and sounds. This multi-dimensional approach is similar to the way humans experience the world. With the proper human training of neural networks and hand-picked quality data sets, DL holds a great deal of promise for powerful, AI-enabled data analysis across industries.
AI can employ both ML and DL algorithmic techniques, but it’s important to remember that AI is much more than that. AI combines specific forms of algorithmic techniques to solve a specific problem or to complete a task that its creators set it up to perform.
Don’t count on one technology: Look for a mix
There is plenty of utility in both ML and DL, but a business shouldn’t attempt to meet all its challenges with just one technology. The real value comes from a tailored combination of these tools. For example, if a retailer has a great deal of traffic data and executives want to extrapolate insight from that data to predict traffic for the holiday season, machine learning will offer valuable insights. However, should that same retailer want to correlate customer traffic data with weather patterns, deep learning would offer a more holistic picture.
The AI services offered by IBM’s Watson can take a lot of the guesswork out of which ML or DL algorithms to use. The engineers of these services have already figured out how to best leverage ML, DL, and other reasoning techniques to make it easy for businesses to build the AI applications best-suited to their particular challenges.
Know your nuance
AI has the power to reinvent the way we do business, but only if company leaders take the time to understand the nuances of these powerful technologies. And while a firm understanding of AI, ML, and DL is a significant step in the right direction, it’s only the beginning of the journey. Decision-makers should take time now to consider how AI can help them fundamentally rethink and optimize their business.
Businesses that think they can wait and see put themselves at a disadvantage. Keep in mind that even if you’re not preparing for AI, your competition is. AI is here, now.
Rob High is an IBM Fellow and vice president and chief technology officer of IBM Watson.
The audio problem: Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here