It’s pretty hard to escape the buzz surrounding artificial intelligence. A June 2017 study by the Mckinsey Global Institute estimates that companies invested in the range of $26 B – $39 B in AI in 2016 alone. A significant portion of this investment has come from tech giants and this massive investment shows no signs of slowing. CB Insights research firm reports that 60 AI startups have been acquired so far in the first two quarters of 2017. Statistics like these are enough to get anyone excited about being involved in pushing this technological frontier, but maybe the only thing harder than escaping the buzz is understanding it.
There’s a reason that everybody is so excited. Artificial Intelligence aims to do exactly what its name implies; it empowers machines with human-like, “intelligent” attributes. Sounds straight forward enough; however, when everyone starts claiming to use “Artificial Intelligence” in their business solutions, the definition of AI becomes harder to nail down.
Before jumping in headfirst to try and identify the true from the fake, it is first critical to understand what the AI landscape looks like and the core problems that AI businesses are trying to solve. At a very high level, you could say that AI as a technology seeks to provide solutions in three general areas:
1. PERCEPTION
2. REASONING/LOGIC
3. ACTION
These three areas represent what has been traditionally seen as “intelligent” attributes.
Whether it be through sight, taste, smell, touch, or sound, as an intelligent human being we have the ability to perceive and make sense of the world around us. This is the first problem that AI solutions tackle. With a rapidly expanding Internet of things (IoT) and innovations in sensor-based technology the way has been paved for machines to develop this same capability. A common example of these perception techniques are virtual assistants like Apple’s Siri, Amazon’s Alexa, or Google’s personal assistant. Using a method called Natural Language Processing (NLP) these virtual assistants can understand when you talk to them and respond accordingly. One major end goal of AI is for machines to be able to sense and operate with a deep and contextual understanding of their environment.
Another area that has previously differentiated human beings from machines is our ability to reason logically and critically. Up until now, machines have been governed by extensive rule based systems that were meticulously written, updated, and edited so that machines would know what to do. Advancements in statistical techniques—aptly named machine learning—have provided a different approach by using supercomputing power to process massive amounts of data. Using this method, data scientists can create software that looks for trends in data and effectively “learns” for itself. The applications for machines who can learn in this manner are endless.
Our ability to physically act for ourselves is something that we have been trying to give to machines for a long time. Previous efforts in robotics have been successful in programing machines to perform specific tasks; however, it’s only recently that the idea of completely autonomous machines has become realizable. The most popular instance of this, of course, is the race to create self-driving cars by tech giants, automobile manufacturers, and startups alike. AI has the potential to enable machines that respond in real time and perform actions that have been previously thought to be too complex.
These three areas—perception, reasoning, and action—broadly represent the overarching goals of the AI movement and illustrate its massive potential for application and growth. It’s evident that the blanket term AI can be used across industries, technologies, and circumstances. By understanding the basic goals of this movement at a high-level, we can save ourselves from getting bogged down in the details and be better suited to make sense of the hype.