ChatGPT is confident and Gemini avoids conflict. Here’s what a workplace personality test reveals about leading AI models.

• The DISC model is a workplace personality test that categorizes behavior into four areas.
• One provider of DISC tested the most popular chatbots, including ChatGPT and Gemini.
• The chatbots seem to have distinct personalities, which could impact workplace dynamics.

OpenAI’s ChatGPT is confident and positive, but when pushed to the extreme, it can be manipulative. Google’s Gemini is a good listener but might need a bit of encouragement to say what it really thinks.

That’s according to DISC, a popular workplace personality test that some companies are now using to help integrate AI models into their workforce.

The test was designed by American psychologist William Moulton Marston, who, back in the 1920s, grappled with fundamental questions about human behavior.

Unlike his contemporaries, Sigmund Freud and Carl Jung, who studied individuals with psychological disorders, Marston studied how people in good mental health interacted with others and their environments. He’s also credited with inventing an early version of the lie detector.

Marston eventually concluded that the human behavior falls into four categories and, in a 1928 book, introduced the DISC personality model. DISC stands for Dominance, Influence, Steadiness, and Conscientiousness, and Marston saw them as the four “primary emotions” that drive human behavior.

Marston’s model has since been formalized into assessments now used by Fortune 500 companies, government agencies, and universities. Questions, which often take the form of statements, might be “I am motivated by accomplishment and authority” or “I am right most of the time,” where participants answer based on how strongly they agree.

Test-takers are ultimately assigned one of 12 types based on their most dominant traits. Someone who receives a D for “Dominance” might be described as direct and strong-willed. Someone who gets a DC for “Dominance Conscientious” might be more meticulous and excel at critical thinking.

As companies accelerate AI adoption, workers are using generative AI to write emails, brainstorm ideas, and conduct research. They might see chatbots like ChatGPT as a competent but neutral sounding board for their thoughts. But these models might not be so neutral.

Online DISC Profile, a provider of DISC, conducted assessments of the most popular AI chatbots. It found that OpenAI’s ChatGPT and Microsoft’s Copilot are both DI, or “Dominance Influence” types. These types are known for being assertive, results-driven, and having a sense of urgency. At their worst, though, they can be manipulative.

Google’s Gemini and China’s DeepSeek are a combination of S, C, and I and can be classified as “steadiness” types, which are rarer. These types are more stable and consistent and tend to be good listeners. They excel at making others feel supported, but they also tend to steer clear of conflict.

Some companies may want to consider how workplace dynamics change when employees spend several hours a week conversing with a supportive but conflict-averse chatbot alongside their teammates.

Managing teams suddenly becomes a bit like managing the cast of characters from Star Wars, according to Sarah Franklin, CEO of Lattice, a platform for managing employee performance and engagement.

“You have Princess Leia leading the rebel command, and she has to get Luke Skywalker, along with R2-D2, the robot, along with Han Solo, along with Chewbacca, and then you have C-3PO, another robot, and like, all of them have to coordinate,” she told Business Insider. “That is very much what we’re doing right now. We have to be able to collaborate together, but you need to have mission control.”

Lattice relies on DISC as an internal assessment to “reduce conflict and improve working relationships.” Some employees have already trained chatbots on their DISC style, Mollie West Duffy, director of learning and development at Lattice, told BI by email.

Duffy said employees can then ask these bots questions like, “I’m a style C, working with a style D on a cross-functional project. How should I give the other person feedback in an effective way, given their style?”

The platform doesn’t leverage work assessment data across its 5,000 customers, which include OpenAI, Anthropic, and E3, but Duffy said it’s something Lattice is considering for the future. Managers might eventually get tailored feedback tips based on their direct reports’ DISC profiles. Or the platform might suggest more personalized growth areas to users.

“We have to have the system where the AI exists as an employee with transparency, accountability, and management,” Franklin said.