AI isn't perfect--but you can get it pretty darn close

3 ways to improve your company's AI accuracy.

The top AI failures of 2017 Artificial intelligence systems are improving rapidly, but when AI stumbles the results can range from humorous to disastrous. TechRepublic's Olivia Krauth shares the top failures of machine learning.

Several years ago, the bank I worked at installed an artificial intelligence (AI) system that detected potential credit card fraud. The system worked well and detected and broke up several fraud attempts.

However, just as we were feeling pretty comfortable with it, we received a phone call from a very irate board member. The board member attempted to make a large purchase at a home improvement store, and his credit card was denied. The situation was inconvenient and embarrassing.

SEE: IT leader's guide to deep learning (Tech Pro Research)

What we learned, unfortunately, was that AI systems are capable of issuing false positives, or inaccurate results. To combat this, we needed to grow the knowledge and analytical capabilities of these systems so systems could move with greater precision.

This process of increasing incremental AI knowledge base and performance should be baked into every AI project. Let's face it: AI isn't perfect. However, if you enact a strong continuous quality improvement process it can get pretty close to it.

How to improve AI's performance

Here are three ways to improve your AI's performance.

1. Choose the right vendor

Interview prospective AI solution vendors about the AI's learning abilities as well as the analytics that the system provides. You also want to know if the AI includes machine learning and deep learning capabilities.

2. Incorporate man-machine business processes

Defining man-machine business processes optimally combine the best that humans and machines can offer each other. For instance, in healthcare AI is adept at quickly processing thousands of pages of medical journals and case histories to come up with a diagnosis for a doctor. The doctor can then refine the diagnosis based upon his/her own empirical years of experience. This results in a best-in-class collaboration between man and machine.

3. Constantly measure results

It's important to constantly measure results. What is the error rate of your AI? What is your accuracy rate? Like other types of software, AI will have shortcomings. The more that you can grow the AI system's knowledge and performance base, the greater the accuracy of results and the return on investment will be.

Also see

istock-968289696ml.jpg
Image: Gorodenkoff Productions OU, Getty Images/iStockphoto