The goal of any Personality AI technology, including Crystal, is to detect the most accurate possible personality profile for anyone, based on the information you have available. This elicits the question, “How can software accurately detect someone’s personality?”
Since every situation is different, there are multiple answers to that question. Sometimes, you can rely on tried-and-true questionnaire-style assessments. When that’s not an option, artificial intelligence makes it possible to analyze someone’s personality from other data like their writing style, social media, career choices, and more. With enough of this data, Crystal can make remarkably accurate predictions about someone’s motivations, behaviors, and communication styles. You can then use these insights to understand others better and communicate more effectively.
To understand how Crystal generates accurate personality insights, you need to understand how it creates two distinct types of profiles: verified and predicted.
Verified profiles come from Crystal users who take our online assessment. The assessment uses a circumplex version of the DISC model to plot out behaviors and personality types on a 360 degree DISC model.
DISC is a four-factor personality framework dating back to the early 1900s that is very popular in management and coaching environments. Within Crystal, hundreds of thousands of people have now used this brief questionnaire to learn more about themselves and their personal communication styles.
To help ensure Crystal’s accuracy, users can read through their own profile or the profile of someone they know, and provide helpful feedback in the form of accuracy ratings of the overall profile, and endorsements of the accuracy of specific personality traits.
Evaluating the correctness of Crystal’s knowledge and communication suggestions is encouraged, as this will help the technology learn and make more precise predictions in the future.
Currently, based on data from thousands of responses, Crystal has a 97% accuracy rating for verified profiles. In cases when someone disagrees with an aspect of their assessment, they are able to retake it and provide specific input.
Predicted profiles are created when you:
These profiles are predicted through machine learning and use text sample analysis and attribute analysis to create a personality profile.
Text sample analysis can detect characteristics of a person’s personality from the style and content of their writing by comparing them to a large data-set of text samples from other people who have validated profiles on Crystal.
Based on comparisons to verified profiles and our user’s direct accuracy validation through ratings and endorsements, Crystal has an 80% accuracy rating for Predicted profiles. This means that when you use Crystal’s Chrome Extension or Dashboard tools to predict a personality, you receive a reliable, precise understanding of roughly every 4 out of 5 people.
In the absence of a large enough text sample, Crystal’s AI uses attribute analysis to look at specific demographic data about a person and make a prediction about their personality type. Each individual aspect, like job title, employer and location can be generally attributed to certain personalities across the population. Crystal normalizes this data and comes up with a “best guess” personality type based on all of the available information.
While our personality assessment used for verified profiles is still the most definitive at an average of a 97% accuracy rating, our Personality AI predictions are not far behind at an 80% accuracy rating. As machine learning technology advances rapidly, Personality AI tools are making it more effective than ever to collaborate and build strong relationships with customers, prospects, coworkers, and friends.
Crystal’s overall goal is to provide the most accurate possible profiles for the largest amount of people, in any situation. By combining methods of machine learning and artificial intelligence, Crystal can deliver accurate personality profiles and insights for virtually any situation.