Throughout 2017, PR practitioners and communicators were bombarded with news, articles, and blog posts about how Artificial Intelligence was moving to the forefront of many businesses. As an industry, we’ve had to keep pace with global changes that affect communications work—whether that means changes within the industry and practice of PR, or external changes that affect client businesses.
The pace of change means that all of this can get a bit overwhelming, and it’s good to occasionally take a quick time-out and catch our breath. To that end, here’s a quick primer on many of the terms being thrown around, because having a solid understanding of these terms is critical to having the ability to see how they are already having an impact on the practice of public relations and communications work—or how they soon will.
“Big Data” is one of those terms that is thrown around so often it’s beginning to be used incorrectly to describe any large data set—which actually diminishes its importance. That said, it can be tough to nail down a precise definition. Doug Laney, a Gartner analyst, defined Big Data back in 2001 (PDF) as having three characteristics: high velocity, high variety, and high volume. Essentially, “Big Data” accumulates quickly, there are a lot of different types, and there is a lot of it.
A year’s worth of clips for a client—no matter how voluminous—probably isn’t “big data.” A year’s worth of customer transactions at Amazon probably is. Clips are one type of data—now, compare that to the components of an Amazon customer transaction, which will contain the following data points (and probably more): how the customer accessed the site, the customer’s browsing history, clicks through to reviews, clicks to review other items, adding items to the shopping cart, and the whole payment transaction process including payment type, selection of delivery method, and ship-to address. There are many different data points gathered during a single customer purchase—now multiply that by all of Amazon’s customers…and you can get a sense of “Big Data.”
Big Data is being used by businesses to tailor recommendations to customers, improve the identification of potential new customers, and much more. It’s being used by municipalities to benefit residents by reducing traffic congestion points to better electricity use, resulting in lower power bills.
Artificial Intelligence (AI)
The most basic definition of AI is that it is machines performing tasks that require intelligence. This definition can include language intelligence, such as asking Siri or Alexa a question, which requires those machines to process the spoken request and provide an answer; or visual intelligence, such as a self-driving car “understanding” that it needs behave a certain way when a street sign, such as a stop sign or yield sign, is encountered.
AI is working its way into all kinds of businesses, who see its value in allowing companies to become more efficient. For example, AI is being used in a number of interactive “chatbots” that are providing customer service or asking survey questions. In these examples, AI is acting in two capacities: it is delivering something (responding to customer service queries or asking survey questions), and it is gathering data (the questions asked in customer service, and the survey responses). Both aspects improve efficiency.
Machine Learning uses AI to “train” computers to make decisions and predictions based on accumulated data. It is considered a component or branch field of artificial intelligence.
The easiest to understand explanation of this process that I’ve heard was articulated by Economist senior editor and author Kenneth Cukier during his TED Talk on Big Data. Cukier described how in the 1950s IBM computer scientist Arthur Samuel liked to play checkers, so he wrote a program to play checkers with his computer. Samuel consistently won these matches, so he wrote a sub-program that ran in the background, allowing the computer to accumulate data on the probability of any given move to result in a winning board versus a losing board. Samuel then left the computer to play against itself—which enabled it to gather a lot of data. More data meant better predictions, which led to the computer winning matches against Samuel. The machine had learned what moves were necessary to win.
Machine Learning is currently being used by Reuters to sift through Twitter data to uncover news stories; Reuters credits the technology with identifying breaking news such as an earthquake in Ecuador, which provided its journalists with lead time allowing them to gather additional information before breaking the story.
Natural Language Processing
Like Machine Learning, Natural Language Processing (NLP) is a branch of artificial intelligence. In NLP, computers “learn” to process human language as it is spoken.
Any PR pro who used automated sentiment analysis when it first started to be widely incorporated into media monitoring software about a decade ago knows and understands the value of NLP. Automated sentiment analysis used to be roughly as accurate as a coin toss, particularly when it was used to score sentiment on blog or social media content. Why? Because sarcasm, snark, and irony were largely missed, and computers had difficulty correctly scoring slang and regional variations in language.
Automated sentiment analysis has improved, and much of the improvement can be attributed to better natural language processing. And, it continues to get better and more sophisticated. Google recently released a study that demonstrated a good degree of accuracy in converting speech-to-text using NLP in a healthcare setting. NLP also helps to make Alexa’s and Siri’s answers more accurate.
How does PR fit in all of this?
The most immediately evident role for the PR practitioner is to know and understand how these aspects of our data-driven lives are changing things right now. While it’s natural to think about the “what-ifs,” the important thing is to pay attention to what is currently driving adoption. In the examples provided above, the “machines” are improving efficiency so that humans can do their jobs better. By offloading repetitive tasks and tasks that can be obscured by biases, machines are allowing humans to make more informed decisions, or freeing up their time so they can perform higher-level tasks.
Creative thinking and effectively synthesizing information that will serve clients’ needs remain the province of humans. That we now have the computing power and storage capacity to process far more data means our decision-making will be better informed—and that’s a vital benefit to effective communications.