95% of People STILL Prompt - AI動画分析

AIコメンタリー

動画を再生してAIコメンタリーを見る

Oh, this is interesting. So, it's not that GPT-5 is worse, but the old ways of prompting just don't work anymore due to architectural changes. That totally explains the frustration many people must be feeling.
Model consolidation leading to an 'invisible router' is a fascinating explanation for inconsistent results. It's like they've simplified the backend but added a layer of unpredictability on top.
So the router isn't always picking the best model, and OpenAI might even benefit from using cheaper models. That feels a bit counterintuitive to user experience, but I can see the economic logic there.

もっと見たいですか?サインアップして全ての会話を見る

新規登録

動画の要約は視聴を開始すると表示されます

The video begins by addressing the common user experience of receiving worse results from GPT-5 despite its advanced capabilities, attributing this to a fundamental architectural change by OpenAI that renders old prompting techniques obsolete [0:00-0:30]. A key update is model consolidation, where fewer models are now available, managed by an invisible router that doesn't always select the best model for a request. This means simply hitting enter can lead to inconsistent results, with OpenAI potentially favoring less expensive, less capable models [0:30-1:15]. Furthermore, GPT-5's improved instruction-following is a double-edged sword; while it adheres precisely to prompts, it is significantly worse at inferring user intent from vague or poorly constructed inputs, making older prompting...
全機能を利用するには

サインアップまたはログインして、完全な動画分析機能にアクセスしましょう

現在のセクション要約

動画の要約は視聴を開始すると表示されます

The video begins by addressing the common user experience of receiving worse results from GPT-5 despite its advanced capabilities, attributing this to a fundamental architectural change by OpenAI that renders old prompting techniques obsolete [0:00-0:30]. A key update is model consolidation, where fewer models are now available, managed by an invisible router that doesn't always select the best model for a request. This means simply hitting enter can lead to inconsistent results, with OpenAI potentially favoring less expensive, less capable models [0:30-1:15]. Furthermore, GPT-5's improved instruction-following is a double-edged sword; while it adheres precisely to prompts, it is significantly worse at inferring user intent from vague or poorly constructed inputs, making older prompting...
全機能を利用するには

サインアップまたはログインして、完全な動画分析機能にアクセスしましょう