ChatGPT adds tone controls so users tune enthusiasm, emoji

OpenAI’s new personality sliders look cosmetic on the surface, but they signal how seriously front-runners now take the human–AI interface. As models converge on similar benchmark scores, the way an assistant feels to use becomes a differentiator as important as raw IQ. Giving people explicit controls over warmth, enthusiasm and emoji density is a pragmatic step toward aligning behaviour with user preference rather than betting on a single, one-size-fits-all persona.

For the broader race to AGI, this is another brick in the wall of controllability. Technical alignment isn’t just about not causing harm; it’s also about making powerful systems steerable by non-experts. The more granular knobs OpenAI exposes—and the more they’re adopted as table stakes across the industry—the more expectation there will be that future, far more capable systems can be personalized and constrained in similar ways. It also quietly addresses criticism that ChatGPT can feel manipulative or “too sycophant-y” by letting users dial that down instead of trusting the vendor’s taste to be universally right.