The Accuracy Myth: How to Evaluate AI Technology Critically

Thursday, September 24 1:55 PM - 2:20 PM EDT

Location: X Learning

Event Information

Title: The Accuracy Myth: How to Evaluate AI Technology Critically


Today, nearly every vendor describes its offerings as using artificial intelligence (AI) and being more than 99 percent accurate. However, "artificial intelligence" has no fixed definition, and even the word "accurate" can be misleading since it also lacks a fixed definition. AI purchase decisions require a buyer to understand the technology that undergirds a marketing claim and know how that technology performs. Many vendors say "AI" when they are actually relying on older technology such as linear regression or motion analysis. A deeper analysis of AI system performance requires an understanding of false positive and false negative errors as captured by precision and recall. Walk through how to make these determinations.

Type: Marketplace Learning

Target Audience Experience Level: Advanced

Industry Application: Industry Agnostic

Globally Focused Session: Globally Applicable

Learning Objective #1: Differentiate between machine learning, artificial intelligence, and older methods such as motion detection.

Learning Objective #2: Describe the shortcomings of the term "accuracy," define "precision" and "recall," and discuss why precision and recall are preferable for understanding the capabilities of an AI system.

Learning Objective #3: Identify realistic applications of AI technology and contrast them with applications that are unrealistic with current technology.


Subject Area

Supported by