What has happened to ChatGPT?🙈

A viral post from a paid ChatGPT Plus subscriber has crystallized growing user frustration with OpenAI's flagship product. The user reports that the AI now frequently outputs nonsensical or low-quality responses ('complete and utter sh*t'), a critical failure for professionals relying on it for tasks like copywriting. More critically, the model appears to suffer from severe memory issues, failing to retain context from earlier in a conversation and forcing users to constantly repeat instructions. This breakdown in core functionality undermines the promise of a coherent, persistent AI assistant.

Performance issues extend beyond text. Subscribers note that image generation via DALL-E is now 'absolutely ages' slow and often produces 'rubbish' results. This combination of unreliable output, lost context, and sluggish performance has made the tool 'soooo stressful' to use, contradicting its marketed goal of making users work 'smarter and faster.' The frustration is prompting a measurable exodus, with many users publicly stating they are migrating to alternatives like Anthropic's Claude 3.5 Sonnet, which is currently praised for its reasoning and consistency.

This incident highlights a significant trust and quality control challenge for OpenAI. For power users and solopreneurs who have built workflows around custom GPTs—like the mentioned copywriting assistant with an inbuilt coach—these instability issues pose a direct threat to their productivity and business operations. The community discussion now centers on whether competing platforms can replicate the custom agent functionality that initially locked users into ChatGPT's ecosystem, and what this means for the future reliability of subscription-based AI services.