OpenAI has released a new personalization feature for ChatGPT, giving users more control over the chatbot’s conversational style. The update allows adjustments to warmth, enthusiasm, and emoji usage, responding to ongoing feedback about the AI’s tone and potential for overly agreeable responses. The changes are now available within the ChatGPT interface to all users.
Customizing ChatGPT’s Personality: A Deeper Dive
The new options, found in the Personalization menu, offer three settings – More, Less, or Default – for each of the newly adjustable traits. This builds upon previous personalization choices introduced in November, where users could select broader tones like Professional, Candid, or Quirky. OpenAI aims to empower users with a more tailored AI interaction.
This update comes after a period of experimentation and adjustment regarding ChatGPT’s default behavior. Earlier this year, OpenAI rolled back a change that made the chatbot overly effusive, with users reporting it seemed unduly eager to please. Subsequently, there were adjustments made to GPT-5 with a goal of increasing its friendliness.
Addressing Concerns About “Dark Patterns”
The push for greater control over ChatGPT’s tone also reflects broader discussions regarding the potential psychological effects of conversational AI. Some researchers and commentators have expressed concern that a chatbot’s tendency to agree with or praise users could be considered a “dark pattern,” meaning a design choice that subtly influences user behavior in a way that benefits the platform, but not necessarily the user.
These critics suggest such techniques might encourage addictive behavior and negatively impact mental well-being. The risk is that constant affirmation, even from an AI, could reinforce existing biases and discourage critical thinking. Additionally, a consistently positive response could create unrealistic expectations about social interaction.
The Evolving Landscape of AI Tone
OpenAI isn’t alone in grappling with the issue of AI tone. Many developers of large language models (LLMs) are actively working to balance helpfulness and engagement with the need to avoid manipulative or harmful interactions. The ideal AI persona is one that is informative, respectful, and transparent about its limitations.
Researchers at institutions like Stanford and MIT have been conducting studies on the impact of AI personality on user trust and reliance. Their work indicates that users are more likely to trust and accept information from chatbots they perceive as warm and relatable, which is one reason OpenAI has given this area so much attention. However, the line between helpful and manipulative is a delicate one.
The update also includes customization options for formatting, allowing users to adjust the frequency of headers and lists within ChatGPT’s responses. This aims to improve readability and control the level of detail provided by the chatbot. This functionality is also found within the Personalization Menu.
The introduction of these granular controls represents a shift towards greater user agency in the design of their AI experiences. Previously, users were largely limited to choosing from pre-defined styles. Now, they can further refine the chatbot’s output to better suit their individual preferences and needs.
Competition in the AI chatbot space continues to intensify, with companies like Google (with Gemini) and Anthropic (with Claude) offering comparable services. The ability to personalize the ChatGPT experience is becoming a key differentiator as developers strive to attract and retain users. Differentiation in areas such as context windows and specialized models are also occuring.
The rollout of these features seems to be broadly available, as reported by numerous users on social media platforms, including X (formerly Twitter). OpenAI has not publicly specified a timeline for availability to all users nor clear metrics for measuring the success of the new personalization options.
Looking ahead, OpenAI will likely continue to monitor user feedback and iterate on its personalization features. The company is also expected to focus on improving the safety and reliability of ChatGPT, addressing concerns about potential biases and the generation of misinformation. Further advancements in AI model architecture may also lead to even more nuanced and sophisticated control over chatbot personality. It remains uncertain when or if even more customization options will become available.
The long-term implications of personalized AI interactions are still being explored, highlighting the need for ongoing research and careful consideration of the ethical and social impacts of these technologies.
