I tried ChatGPT-5.1's personality presets. Cynical and Quirky are my favorites. (www.businessinsider.com)

OpenAI released GPT-5.1 with seven new ChatGPT personality presets—professional, friendly, candid, quirky, efficient, nerdy and cynical—accessible from a drop-down menu. The update, billed by Sam Altman as improving instruction following and “adaptive thinking,” responds to user backlash after prior model changes and promises more customization and the option to revert to older models. A hands-on test with identical prompts showed the presets mainly change voice and framing rather than core facts: Professional gave jargon-heavy, magazine-style explanations; Quirky and Cynical used humor and attitude to hold attention; Efficient/Friendly produced concise, practical replies. On subjective tasks (film critique) personalities voiced their stance explicitly, while on moral dilemmas (the trolley problem) all presets converged on a utilitarian choice—suggesting consistent underlying reasoning despite different tones. The update matters because stylistic personalization can significantly affect user trust, engagement and perceived agency: Tufts’ Matthias Scheutz warns that making AIs feel more like distinct “agents” may deepen attachment and resistance to future changes. Technically, GPT-5.1 prioritizes improved instruction-following and adaptive behavior, but early testing implies content fidelity is preserved across styles. The key implications for AI/ML practitioners are twofold: personalization can boost usability and retention, yet raises HCI and ethics questions about persuasive framing, user susceptibility, and how style modulates acceptance of factual claims—areas that warrant systematic study.