OpenAI allows users to directly adjust ChatGPT’s enthusiasm level

In Brief

ChatGPT users tin present tweak the chatbot’s warmth, enthusiasm, and emoji use, according to a societal media station from OpenAI.

These options (as good arsenic akin adjustments to ChatGPT’s usage of headers and lists) present look successful the Personalization paper and tin beryllium acceptable to More, Less, oregon Default. They let users to further customize ChatGPT’s tone, connected apical of the existing quality to acceptable a “base benignant and tone” — including the Professional, Candid, and Quirky tones that OpenAI added successful November.

ChatGPT’s code has been an ongoing contented this year, with OpenAI rolling backmost 1 update for being “too sycophant-y,” past aboriginal adjusting GPT-5 to beryllium “warmer and friendlier” aft immoderate users complained that the caller exemplary was colder and little friendly.

Some academics and AI critics person suggested that chatbots’ inclination to praise users and affirm their beliefs are a “dark pattern” that creates addictive behavior and tin person a antagonistic effect connected users’ intelligence health.

Subscribe for the industry’s biggest tech news