Archive — US — techcrunch
GPT-5 is supposed to be nicer now
OpenAI announced an update for GPT-5 aimed at making it 'warmer and friendlier' after a bumpy launch. Users preferred the prior GPT-4 model, prompting changes for improved approachability, including subtle expressions like 'Good question'. OpenAI executives acknowledged the rocky debut at a recent dinner with journalists.
Key points
- OpenAI launched GPT-5 recently, which faced user criticism and complaints.
- CEO Sam Altman admitted the launch process was bumpier than anticipated.
- Users expressed a preference for GPT-4 over the new model.
- The update aims to enhance GPT-5's personality with touches like 'Good question,' without increasing sycophancy.
- OpenAI executives addressed the model's rocky start during discussions with journalists.
Why it matters
- This update aligns with current trends focusing on user experience and AI interaction improvements.
- The push for more human-like AI responsiveness is part of a broader trend in 2025 IT developments towards AI being more intuitive and user-friendly.
- Historically, feedback from initial AI releases has often led to rapid iterations and improvements based on user preferences.
OpenAI's effort to update GPT-5 reflects a commitment to user satisfaction, addressing feedback head-on to enhance engagement and approachability in AI, a crucial aspect of the evolving tech landscape.