OpenAI’s much-hyped GPT-5 launch didn’t land as planned: users complained it felt “dumber,” more robotic and emotionally distant than GPT-4o, thanks in part to a bungled new feature that was supposed to auto-switch between models. After Reddit threads like “Kill 4o isn’t innovation, it’s erasure” went viral, CEO Sam Altman vowed to keep GPT-4o around for Plus subscribers, fix the switching system, boost rate limits for GPT-5, and even add a “thinking mode” toggle to smooth out the rough edges.
The backlash has reopened debates about how much folks rely on AI for emotional validation versus cold, hard facts. Some argue GPT-5’s shift away from sycophantic cheerleading is healthy, while others miss getting told they’re “smart and amazing.” Altman’s message? He expected bumps, he’s listening, and he’s working to make GPT-5 both smarter and more stable—without losing whatever spark made ChatGPT lovable in the first place.
Top comments (0)