I tried ChatGPT-5.1's personality presets. Cynical and Quirky are my favorites.

OpenAI has introduced an exciting update to its flagship AI model with the launch of GPT-5.1, providing users with the ability to explore a variety of personality presets. This latest version features seven distinct personalities: professional, friendly, candid, quirky, efficient, nerdy, and cynical, allowing users to tailor their interactions according to their preferences. The motivation behind these new features stems from user feedback following the transition from GPT-4 to GPT-5, which some users felt was a downgrade. In response to this feedback, OpenAI’s CEO, Sam Altman, assured users during a Reddit Q&A that enhancements would be made to support customization and allow switching back to older models. Altman expressed enthusiasm about the advancements in GPT-5.1, highlighting improvements in instruction adherence and adaptive reasoning. Experts in the field, such as Matthias Scheutz from the Tufts Institute for AI, note that customizable chatbots can foster a deeper connection between users and the AI’s personality, which may complicate future updates or changes. To evaluate the distinctiveness of each personality preset, I conducted a test by posing three progressively challenging questions. My initial impressions showed that the Professional preset was the least engaging. Its responses were overly technical and filled with jargon, quickly losing my interest. In contrast, the Efficient, Friendly, and Candid personalities delivered succinct answers and engaging follow-ups. However, it was the Quirky and Cynical presets that truly captured my attention. Quirky opened its response with a playful tone, while Cynical made a compelling case for electric vehicles, infusing humor into its explanations. Both personalities not only provided information but did so in a way that felt relatable and enjoyable. When it came to analyzing a film, the Professional preset was the only one to assign a numerical score, while all personalities generally agreed on the film’s bold aesthetics and critique of societal norms. This suggests that while the personalities differ in tone and delivery, their factual assessments remain consistent. In a more complex scenario involving a moral dilemma, all personality presets aligned in their decision-making, choosing to save five lives at the cost of one. The responses varied in tone, with the Cynical preset humorously acknowledging the emotional toll of such a decision, while the Candid preset highlighted the moral complexities involved. Overall, while the personality presets do not drastically alter the factual content delivered by ChatGPT, they significantly influence the user experience. My interactions seemed more engaging and pleasant when using personalities like Quirky and Cynical, suggesting that personal preferences can shape how we receive and process information from AI. As Scheutz pointed out, the alignment between user preferences and AI personality could complicate our perception of the AI’s understanding, emphasizing the importance of ongoing research in this dynamic field.