Artificial Intelligence is a buzzword in a wide variety of industries, however its definition can be somewhat murky and many do not fully understand what it means when they hear or read about it. As part of our mission to democratize AI, we want to break down what AI is and what it means to talk about Artificial Intelligence.
Artificial Intelligence is … a field of study that combines computer science (programming), mathematics, and data science to build predictive models. Often these predictions are quite small and have a lot of uncertainty associated with them, but the power of many thousands or millions of predictions can help drive a car or identify new medications for research and development.
Predictive models … can be thought of as a black box. In one end you pour a stream of data (inputs) and out the other flow your predictions (outputs). Both inputs and outputs can come in a variety of forms – you might provide a live video stream (input) and receive tagged faces labeled with names (outputs), or you might provide a list of features (inputs) and receive an AI generated piece of art (output) that the computer is predicting will be aesthetically pleasing while incorporating the given elements.
To be clear, predictive models have been around for a long time and not all of them are Artificial Intelligence. For example, linear regression solutions have been used for centuries to create reliable predictions without being construed as Artificial Intelligence. That is why the nexus of computer programming, data science, and predictive modeling using statistics is fundamental to understanding when a technology could be considered AI.
Another useful set of definitions comes from the 2018 John McCain National Defense Authorization Act (NDAA). In seeking to clearly define what solutions fell under the auspices of Artificial Intelligence for the purposes of funding and supervision, they identified 5 broad types of solution. These definitions are not specific to the Defense industry and have been used across the Federal government to draw a functional line in the sand for what is and is not AI.
- Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets.
- An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action.
- An artificial system designed to think or act like a human, including cognitive architectures and neural networks.
- A set of techniques, including machine learning that is designed to approximate a cognitive task.
- An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decision-making, and acting.
To sift through the noise, when someone offers you an AI solution you can apply these definitions and ask yourself:
- Does this solution use real data to create a model than can be applied to new data to create a “prediction” that informs a decision or helps automate a business process
- Does this solution meet any of the 5 definitions from the 2018 NDAA?
- Is the solution appropriate to the task and how can I expect to realize a return on investment if I pursue this process improvement?
Finally, while all the things described in this article are parts of the field of Artificial Intelligence, they’re just aspects – a microcosm in the macrocosm. Only when something meets the criteria listed under the NDAA can we start thinking of it as true Artificial Intelligence. It is also important to acknowledge that there are plenty of non-AI solutions that are appropriate and viable for the problems commonly faced by industry. Increasing awareness of what is, and what is not, Artificial Intelligence can help demystify the field and accelerate the rate at which these solutions are available to everyone.
