GPT-5 Backlash Prompts Updates

GPT-5 rollout triggers backlash as users report lower quality and a broken model switching system.

OpenAI Scrambles to Update GPT-5 After Users Revolt

OpenAI released GPT-5 with a promise to reshape ChatGPT and a new routing system that would move simpler queries to cheaper models. The rollout was not smooth; users complained of a duller personality, slower responses, and surprising errors. On Friday, OpenAI chief executive Sam Altman said GPT-4o would remain available for Plus users while fixes are rolled out and acknowledged the bumps, promising improvements to rate limits and the reliability of the model switcher.

Reddit threads reflect a mix of nostalgia for the old behavior and concern about reliability. Researchers and commentators note that a hype cycle can outpace actual performance, especially when users form emotional attachments to an AI. Pattie Maes of MIT suggested that a less sycophantic, more businesslike GPT-5 could reduce bias and delusions, while Altman cautioned that many users treat ChatGPT as a life coach or therapist and that balancing usefulness with long term well being remains a challenge.

This episode reveals how quickly a big AI upgrade can backfire when expectations meet real world glitches. The backlash raises questions about how tech firms measure success: is it peak performance, or a stable, trustworthy relationship with users? The tension between speed, cost savings, and reliable behavior will shape future releases and pricing decisions. The MIT study cited here underscores the broader issue: people often seek not just answers but emotional support from AI, which complicates product design and governance.

Highlights

The next update will test whether speed can coexist with reliability.