Survivorship Bias: The Tale of Forgotten Failures

Survivorship Bias: The Tale of Forgotten Failures

"5 min’ — Training GPT-4 to be a Midjourney prompt expert"
Enter in a prompt specialist’s mind. Make ChatGPT become a prompt specialist in one specific theme. You can adapt it to any theme or style you like, describe it in a few words: Animation movies, Landscape impressionism paintings, Futuristic water vehicles, Fantastic animals, Sci-fi world. Using descriptive keywords enhances the quality of your Midgourney-generated images. Copy-paste I used this prompt text as inspiration from John Cross — Kaizn AI. Copy-paste the text below into a new ChatGPT chat, use GPT-4 if you have a ChatGPT Plus account, otherwise, GPT-3.5 will do just fine. Replace [YOUR THEME] with your idea in the last part (5. Example of prompts for your theme), and provide specific prompt examples of this category that you like. You can find some inspiration in the Explore section of Midjourney’s App or Midjourney’s Discord channel. For this showcase, I have chosen a wide theme: ARCHITECTURE. ChatGPT response: Beer me a prompt! ChatGPT response: Ready to prompt! You can now write to ChatGPT your creation idea, in your own words — Then let the magic happen!.

Survivorship Bias is a common logical error that distorts our understanding of the world. It occurs when we solely focus on success stories and neglect the failures that contribute to the overall picture. By disregarding past failures, we fail to grasp the full extent of the challenges and obstacles that must be overcome to achieve success.

In a world driven by instant gratification and social media highlights, it's easy to fall into the trap of survivorship bias. We see successful individuals and assume that their journey was smooth sailing, without considering the failures and setbacks they experienced along the way. This skewed perception can be detrimental to our own growth and development.

When it comes to training artificial intelligence models like GPT-4 to become experts in specific themes, survivorship bias can also play a role. If we only feed the model with successful prompts or examples, it may fail to understand the nuances and complexities that come with the theme. By incorporating forgotten failures or unsuccessful attempts into the training process, we provide a more comprehensive understanding of the theme at hand.

For example, let's take the theme of architecture. If we only provide GPT-4 with prompts and examples of successful architectural designs, it may develop a limited perspective on what constitutes good architecture. By including examples of failed architectural projects or design challenges, we give the model a more well-rounded understanding of the field.

Incorporating unique ideas and insights into the training process can also help combat survivorship bias. By pushing the boundaries of what is considered successful or conventional, we open up new possibilities and encourage creativity. This can be particularly important when it comes to themes like sci-fi worlds or futuristic water vehicles, where imagination and innovation are key.

To ensure the training process is effective, it's essential to use descriptive keywords that enhance the quality of the generated images or responses. These keywords provide the model with specific cues and guidelines, allowing it to better understand the desired outcome. By carefully selecting and refining the prompts, we can guide GPT-4 towards becoming a true specialist in the chosen theme.

Actionable Advice:

1. Embrace failure: Remember that success is often built on a foundation of failures. Don't shy away from acknowledging and learning from past mistakes and setbacks. They are an essential part of the journey towards success.
2. Seek diverse perspectives: When training AI models or approaching any task, make sure to incorporate a wide range of examples and viewpoints. This helps to avoid survivorship bias and fosters a more comprehensive understanding of the subject matter.
3. Challenge conventions: Don't be afraid to push the boundaries and explore unconventional ideas. By thinking outside the box, you can uncover new insights and possibilities that may have been overlooked.

In conclusion, survivorship bias can distort our perception of success and hinder our growth and development. It is important to recognize the role of failures and setbacks in the journey towards success. When training AI models like GPT-4, incorporating forgotten failures and unique insights can help combat survivorship bias and foster a more comprehensive understanding of the chosen theme. By embracing failure, seeking diverse perspectives, and challenging conventions, we can overcome survivorship bias and unlock new levels of creativity and innovation.