95% of People STILL Prompt ChatGPT-5 Wrong

About this video

• Video Title: 95% of People STILL Prompt ChatGPT-5 Wrong
• Channel: Jeff Su
• Speakers: Jeff Su
• Duration: 00:11:51

Overview

This video explains why users are experiencing worse results with ChatGPT-5 despite its advancements. The presenter, Jeff Su, details fundamental changes in GPT-5's architecture that make old prompting techniques less effective. He then offers five tips to improve prompt quality and achieve better outputs from GPT-5.

Key takeaways

• Fundamental Changes in GPT-5: OpenAI's GPT-5 architecture has changed, making previous prompting methods less effective.
• Model Consolidation and Routing: GPT-5 now uses a consolidated model structure with an invisible router. This router may not always select the optimal model for a request, potentially leading to worse results.
• Improved Instruction Following: GPT-5 is better at following instructions, which is beneficial for complex tasks but makes it less forgiving of vague or poorly constructed prompts.
• Tip 1: Router Nudge Phrases: Adding specific phrases like "Think hard about this" can encourage GPT-5 to use higher reasoning models, leading to more nuanced and insightful responses.
• Tip 2: Verbosity Control: Using phrases like "give me the bottom line in 100 words or less" or "aim for concise 3 to five paragraph explanation" helps control the length and detail of GPT-5's output.
• Tip 3: OpenAI's Prompt Optimizer: This tool can rewrite prompts to be more effective for GPT-5, though it requires a developer account and payment method. A free workaround using a meta-prompt is also provided.
• Tip 4: XML Sandwich: Structuring prompts with XML tags to clearly label different sections like context, task, and output format improves GPT-5's comprehension and the quality of its responses.
• Tip 5: The Perfection Loop: Instructing GPT-5 to develop its own rubric, self-critique, and iterate until it meets a high standard, especially for complex tasks, can lead to superior results.