So You Want to Train a 13-Year-Old Genius: Getting ChatGPT Set Up Right

Article content

Step 1: Open Your Paid ChatGPT Account and Navigate to Settings

  1. Log in to ChatGPT and click on your profile picture (Top Right Corner) – This opens up settings and customization options.
  2. Select ‘Settings’ – This is where you’ll configure how ChatGPT behaves.
Article content

Step 2: Enable Memory for Smarter Responses

  1. In ‘Settings,’ scroll to ‘Personalization.’
  2. Ensure ‘Memory’ is ON.
  3. Click ‘Manage Memory’ to see what ChatGPT remembers about you. If there’s anything incorrect or unnecessary, click the garbage can icon to remove specific memories.
Article content

Step 3: Customize ChatGPT’s Instructions to Fit Your Style

  1. In ‘Settings,’ go to Personalization, go to ‘Custom Instructions.’
  2. Fill in the following fields:
  • ‘What should ChatGPT call you?’ – Give it a name you want it to call you.
  • ‘What do you do?’ – Describe what you do to help shape its answers better.
  • ‘What traits should ChatGPT have?’ – Define how it should interact with you.


5. Under ‘Advanced Settings,’ ensure all options are checked.

6. At the bottom, turn on ‘Enable for new chats.’

Pro Tip: This is like telling your GPS that you hate highways—it won’t put you on one unless absolutely necessary.

Step 4: Choose The Right ChatGPT Model (Because Not All Are Equal)

Not all versions of ChatGPT are created equal. If you’re doing writing or any structured work, you want the best engine available—which means using GPT-4o.

  1. In the chat window, look at the model selector at the top.
  2. Select ‘GPT-4o’ for the best quality responses.
  3. Avoid lighter models unless needed:
  • GPT-3.5 – Faster, but less nuanced and structured.
  • o1 – Best for quick, low-priority tasks where accuracy isn’t critical, such as brainstorming rough ideas or summarizing casual conversations.
  • o3-mini – A lightweight model designed for speed over depth, useful for simple back-and-forth queries or when rapid responses are needed.
  • GPT-4 Turbo (if available) – A balance between speed and quality, but not always better than 4o.
  • GPT-4o with Scheduled Tasks, GPT-4.5 Research Proview, o3-mini-high are all new models recently released.

Why: GPT-4o gives more accurate, structured responses. GPT-3.5 is like asking a 7-year-old instead of a 13-year-old—it’ll try its best, but it might color outside the lines. Lighter models like o1 and o3-mini are great for quick, informal interactions but should be avoided for detailed, structured work like resume editing, proposal writing, or precise business documents.

Step 5: Test Your Setup

Now that everything’s configured, let’s run a test. Start a new chat and ask ‘What do you remember about me?’ It should respond with some of your preferences. Want to try something fun? If you have been using ChatGPT for a while and have items in memory, ask ChatGPT to humourously roast you!

Article content

Final Thoughts: Why This Setup Matters

With ChatGPT, the more specific you are at the start, the less frustration you’ll have down the road. Now that we’ve set up your AI assistant, in the next article, we’ll cover how to actually talk to it without wanting to pull your hair out.

Leave a Comment