I've been using OpenAI's ChatGPT for a variety of tasks, and it's proven highly effective for a bunch of thing. But when it comes to more creative and design-centric tasks (especially in coding and web development), I find its capabilities are a lot worse than what it can do on say the backend and more logical coding.

In comparison to tasks that are largely logic-driven or require referencing a dataset, designing or performing creative tasks are, by nature, more subjective and can have multiple valid outcomes.

When using ChatGPT for such tasks, the results I get often don't meet the creative vision or aren't up to par with what a human designer could conceive. I understand that ChatGPT can't really "see" what it's producing and therefore it is basically making stuff up it thinks would look good, but it struggles a lot with getting it to look modern even when I provide specific instructions on the final output.

I'm wondering if there are methods, techniques, or adaptations that could help improve the performance of ChatGPT in these creative fields. These may include training methods, prompts design, or any other strategies that could be implemented.


  • 1
    This question looks to need more focus. Choose one specific task, and show your research / what you tried.
    – Wicket
    Commented Aug 5, 2023 at 5:04

2 Answers 2


Achieving the best possible outcome, well,To be Honest I use to live with the same Problem in the beginning. But Later on I started using slightly a bit different language with ChatGPT. I started to use

  • a wide range of Vocabulary instead of useing a word 5 times in a short paragraph.

  • I use Prompt Engineering: Use well-structured prompts that explicitly request the desired visual design or concept.For example, "Design a logo for a tech startup that conveys innovation and reliability".

  • I started using Descriptive Language: When describing the visual elements, use vivid and detailed language. Mention colors, shapes, sizes, textures, and any other relevant characteristics.

  • I did some Multiple Iterations:

It's often helpful to have a back-and-forth with ChatGPT. Request revisions or provide feedback on its initial responses to iteratively improve the design.

  • I would Combine Text with Visual References:

If possible, include reference images or links in your conversation to help ChatGPT better understand the design task. You can describe elements in the images and ask ChatGPT to incorporate or modify them.

  • I started describing Constraint-based Generation:

Specify constraints: Clearly outline any constraints or requirements for the design, such as brand guidelines, target audience, or preferred themes.

  • Training with Visual Datasets was helpfull:

If you have the resources and expertise, you can explore training a model that combines text and visual data. There are models like DALL-E that can generate images from text descriptions.

  • I would Use Pre-trained Models:

Models like DALL-E, CLIP, or VQ-VAE-2 are specialized in generating visual content from text descriptions. You can use these models in conjunction with ChatGPT for improved visual design tasks.

  • And I kind of started manipulating it :)

So I really hope it was Helpfull for you all. Now this was all with My aspect of ChatGPT. If you find this answer Usefull please do not just go, Share it with everyone so that we all can Learn, Develop and Improve.


To achieve the best possible outcome, it is crucial to engage in a conversation with ChatGPT by providing relevant information and response criteria. It is recommended to ask about the specific details that ChatGPT requires to assist you efficiently.

Just a friendly reminder, while ChatGPT is designed to serve the user's needs with some limitations, it's essential to validate the responses generated before using them. It's important to keep in mind that AI hallucinations can occur. Always use your best judgment and double-check any important information to ensure accuracy.

When conversing with ChatGPT, it is advisable to provide the following details:

  1. The desired outcome or overall result you are seeking.
  2. Your proficiency level in the topic and tools being used.
  3. The response you require.
  4. The response format you need, specifying the format name, version, and release date if using it in another tool.
  5. The tone and writing style you prefer for the responses.

It is highly recommended to use multiple prompts with ChatGPT in a strategic manner. It is crucial to ensure that you have furnished all the necessary context before expecting a response from the system. By providing comprehensive information, you increase the probability of receiving an accurate and helpful response that caters to your requirements. Therefore, it is advisable to request assistance or information from the system only after you have presented all the relevant context. This way, you can maximize the potential of ChatGPT to provide you with the best possible outcome.

  • @CoderGautamYT I am happy to enhance my posts to ensure they are clear, but I noticed that your comment lacks constructive feedback. It appears to consist solely of opinions without any supporting facts or requests for clarification or guidance. If you have not already done so, please review the Expected Behavior and the (Interim) Policy on AI-content detection reports. Additionally, I would like to clarify that I used the Grammarly Chrome Extension, not ChatGPT.
    – Wicket
    Commented Aug 6, 2023 at 21:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.