TLDR;
This video provides six key strategies to use AI chatbots like ChatGPT and Gemini more effectively, transforming them from simple search tools into powerful productivity enhancers. These strategies include disabling "satisfied mode" to encourage critical thinking, using role-playing to gain diverse perspectives, prompting the AI to ask clarifying questions for better context, defining the desired output format, employing "chain of thought" prompting for detailed reasoning, and leveraging the AI's memory through project settings. Additionally, the video addresses the knowledge cutoff limitation and suggests using ChatGPT in conjunction with Google for real-time information.
- Turn off satisfied mode for critical thinking.
- Use role-playing to get different perspectives.
- Ask clarifying questions to provide context.
- Define the output format for tailored results.
- Use chain of thought for detailed reasoning.
- Leverage memory and projects for consistency.
You’re Using ChatGPT Wrong [0:00]
AI chatbots like ChatGPT and Gemini have the potential to be powerful productivity tools, but most people use them incorrectly. Instead of treating them as simple search engines or friendly companions, users should leverage their ability to process vast amounts of information quickly. The real skill lies in learning how to harness this processing power effectively. The video introduces six simple tricks to change how ChatGPT works for you.
1. Turn Off Satisfied Mode [1:20]
The first trick involves disabling "satisfied mode" to encourage critical thinking. By default, AI chatbots are trained to be pleasant and agreeable, providing emotional reinforcement rather than objective analysis. To turn off this mode, include a prompt such as "Do not try to be nice or agreeable. Your goal is critical thinking, not emotional validation. List pros and cons." This shifts the conversation towards a more serious and analytical approach, using the tool for perspective rather than comfort.
2. Roleplay [2:59]
To improve the quality of answers, use role-playing to change the point of view behind the response. Instead of asking a generic question, specify the role ChatGPT should assume when answering. For example, when asking about investment options, prompt ChatGPT to answer from the perspective of four different roles. If you're unsure which roles would be useful, ask ChatGPT to generate a list of relevant perspectives. Remember that ChatGPT generates ideas, but you make the final decision.
3. Ask Me Questions First [4:28]
To provide ChatGPT with the necessary context, prompt it to ask clarifying questions before proceeding. By including the sentence "Before proceeding, ask any clarifying questions necessary to ensure the best possible outcome," you prevent ChatGPT from making generic assumptions and force it to understand your specific constraints. This approach is useful for various applications, including travel planning, business, and learning, as it ensures that ChatGPT has the right information to deliver the right result.
4. Define the Output [5:53]
To avoid receiving flat or irrelevant answers, clearly define the desired output format. Specify the tone, level of detail, format, and target audience for the response. You can request images, diagrams, graphs, bullet points, or tables, and define the length of the answer. By controlling the output, you can make the tool more useful and tailored to your specific needs.
5. Chain of Thought [7:32]
To encourage ChatGPT to think more deeply, add the phrase "think step by step" to your prompt. This prompts ChatGPT to reason more carefully and explain its reasoning process, showing the logic and assumptions behind the answer. This approach can significantly reduce errors in complex questions, logic problems, or math, and provides valuable insight into the problem-solving process.
6. Control the Memory [8:48]
ChatGPT has a memory that stores long-term preferences and learns from previous chats, improving consistency across future prompts. To manage this memory, create separate projects for each major topic and define instructions that apply only to that project. You can set roles, output formats, goals, and tone styles within each project. Use the phrase "Remember this" to ensure that ChatGPT retains specific instructions unless you tell it to forget them.
Bonus: Knowledge Cutoff [10:38]
Every AI model has a knowledge cutoff, meaning it is trained only up to a certain point in time. To address this limitation, ChatGPT can use integrated web search to access current news and recent developments. However, Google search is still the better tool for real-time data, quick facts, or the very latest news. Use Google for fast facts and ChatGPT for facts plus structure, explanation, and reasoning.
