How to Prompt Claude and ChatGPT effectively | Good vs Bad examples

Value of the Information & Strength of the Argument

The video provides high practical value by offering concrete, actionable examples that viewers can directly apply. The argumentation is solid, as each example clearly demonstrates the difference between vague and specific prompts, and the reasoning behind why the improved prompts yield better results is logical and well-explained. The author avoids overcomplicating the topic and focuses on simple, effective techniques. The emphasis on treating AI as a colleague needing context and iteration is a useful mental model. However, the video lacks empirical evidence or references to support the claims, relying instead on anecdotal experience. The argumentation is persuasive but not scientifically rigorous.

Scientific Rigor, Source Quality, Title Accuracy

The video does not cite any external sources, which limits its scientific rigor. The only link provided is to the author’s website, which is not a source for the content. The title accurately reflects the content, and the video stays on topic throughout. The lack of citations is a significant weakness for a video claiming to be a ‘masterclass’ in prompting. The advice given is generally consistent with common knowledge in the field, but without references, it is difficult to verify the accuracy of the claims. The video’s strength lies in its practical examples rather than its scientific foundation.

The title accurately reflects the content, which focuses on effective prompting with examples of good and bad prompts for ChatGPT and Claude.

6/10

The video provides practical, actionable advice on prompt engineering, but lacks citations to scientific studies or authoritative sources. The content is based on the author's experience and common practices, which are generally consistent with known principles, but the lack of verifiable references reduces its reliability.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction: Most people have a prompting problem, not an AI problem.
• 01:00Mental model: AI is like a capable colleague with zero context; needs context, specifics, examples, iteration.
• 02:30Example 1: No context prompt - 'Write a blog post about productivity' vs. detailed prompt.
• 04:00Example 2: 'Make it better' - vague feedback vs. specific notes.
• 05:30Example 3: Polite but useless - 'Hi, can you help me with my resume?' vs. front-loading task.
• 07:00Example 4: Everything prompt - 'Help me start a business' - one task at a time.
• 08:30Example 5: Format specification - 'Compare these tools' - ask for table.
• 10:00Example 6: Brand voice - paste examples instead of describing style.
• 11:30Example 7: Summarize for a specific reader and decision.
• 13:00Example 8: Iteration - treat first draft as draft, give specific notes.
• 14:30Example 9: Role prompting - 'Act as skeptical consultant' - but specificity still matters.
• 16:00Example 10: Avoid leading questions - ask for strongest case against.
• 17:30Example 11: Data analysis - provide hypothesis and ask to flag undercutting evidence.
• 19:00Example 12: Set standing rules at the top of conversation.
• 20:30Example 13: Verify facts - ask what AI is least confident about.
• 22:00Conclusion: Six-part template (Role, Context, Task, Format, Example, Guardrail) and encouragement to apply.

• Dr. Asif's Mol. Biology website — The only link provided in the description, likely for courses and additional resources.

• OpenAI Prompt Engineering Guide — Official guide that aligns with the video's advice on specificity and context.
• Anthropic's Prompt Engineering — Official guide for Claude, consistent with the video's recommendations.