Is 2018 The Year Your Business Needs To Prep For AI?

Just a few years ago, it might have been easy to brush aside artificial intelligence (AI) as a fantasy more suited to science-fiction novels than everyday life. Today, that would be foolish — anyone who owns an Amazon Echo or talks to Siri already interacts with AI on a daily basis, and this is only the beginning of a long-term trend.

If you’re a business manager, you can’t afford to ignore this.

As I’ll outline below, artificial intelligence will soon have quite a few practical (and near-term) applications for business. However, it won’t be something you’ll be able to decide on one day and implement the next. At least not if you’re unprepared, that is. Instead, your business must make specific preparations now — even if the AI-based applications you end up using are still a couple of years off.

Before I explain, there are some basic concepts about AI you’ll want to understand that will help you see how using AI is unlike a typical software as a service (SaaS) experience.

What Exactly Is Artificial Intelligence Anyway?

Artificial intelligence refers to technologies such as machine learning, deep learning, neural networks and natural language processing.

Instead of explaining each of those technologies, for the purposes of this article I’ll simply define AI as software that can learn from historical data and continually improve itself using new data. What’s important is that AI is a break from traditional logic-based computing, where a computer is given a specific (and fixed) set of instructions.

Consider a programmer designing a self-driving car. A typical logic-based approach would involve telling the car, “If this scenario happens, then do this”. A programmer would need to write instructions for each of the millions of scenarios the car would encounter. This is an all but impossible task. The AI-based car, however, would be able to learn from experience and revise its own settings and parameters. The result is that the AI-enabled car will, over time, become much more proficient at driving than its logic-based counterpart ever could.

This is just one example. The point is that artificial intelligence allows us to solve entirely new problems. This is especially true for those where massive amounts of data or complex sets of rules would apply.

Class Is Almost Over, But Make Sure You Know These Terms

There are just a few more terms you’ll want to know before we move on.

First, AI learns through a supervised or unsupervised processes. Supervised means that the software receives “teaching data” — information that is already classified (think pictures of cats). Once AI software receives initial set of data, a “supervisor” can monitor its responses to new data and correct the software along the way.

Unsupervised learning is the opposite. AI software receives unclassified data, and it’s up to the software to find patterns on its own. This approach is much more complex (and uncommon).

Finally, AI is either applied or general. Applied (sometimes called weak or narrow AI) means that the software was developed for a specific purpose (like our self-driving car). General (or strong) AI can theoretically do anything. It’s the kind of technology that comes to mind when thinking of an android (or Skynet).

When your business makes its first use of AI, it’s likely going to be the supervised, applied sort.

For Many Businesses, AI Will First Help With Prediction

There are myriad ways AI can be used in business, but one of the most straightforward is in predictive analytics. AI excels at reducing the cost of making predictions. Because nearly every business needs to make predictions about something, I think this application of AI stands to become commonplace. Here are a few examples of what it could do:

  • Predict which customers are likely to churn
  • Forecast pricing chances of competitors
  • Understand weather’s effect on in-store sales
  • Suggest content for website visitors to read
  • Recommend items based on past purchase history

You’ve probably noticed something — all of these examples would require lots of historical data for initial predictions to be accurate. That brings us to the crux of this post. To implement AI, your business will need to get its data in order well ahead of time.

Given Poor Data, AI Will Make More Decisions

The value your business gets out of AI is a function of the quality and quantity of data you provide it. That’s why your business needs to start preparing now. The more clean, classified and meaningful data you can gather now, the more you’ll get out of AI in the future.

You don’t need some grand plan to restructure all your data all at once. Instead, start by picking one area of your business that could be improved with better predictive abilities. Then, take a look at the data you are collecting, and see how it measures up against these criteria:

  1. It’s recent. Do you have a way to quickly gather and store new data? If not, consider how you can automate and expedite current data collection processes. One of AI’s great advantages is that it can improve predictions with new data. But that’s not very helpful if it’s acting on outdated information.
  2. It’s structured. However your company implements AI, you’ll need data that are consistently formatted and labeled. Start putting the right structure in place now so don’t have a huge mess to clean up later.
  3. It’s sufficient. Any prediction that’s based on a small set of data will likely be a poor prediction. If you’re not collecting enough data, even the best AI technology won’t be of much use to use to you.

You Can’t Afford To Be Unprepared for AI

While blockbuster movies tend to portray AI as a dystopia-inducing threat to mankind, reality will probably something more mundane yet highly practical. Few businesses don’t stand to benefit from making better predictions about the future, and if you are intentional about preparing now, you’ll be in a good position to turn artificial intelligence into a competitive advantage.


This post was originally published in Forbes.

Leave a Reply