Crafting Your Own AI: How to use the GPT Store to build your custom GPT
OpenAI provides an app store-like platform for publishing and sharing GPTs, custom applications that take advantage of the underlying OpenAI infrastructure.
GPTs - OpenAI's customizable ChatGPTs
GPTs are customized versions of ChatGPT that you can customize for your specific purpose. GPTs allow non-technical users to create their own customized version of ChatGPT for specific tasks.
The GPT store allows you to share your custom GPT with others or use someone else’s GPT.
How to create a GPT - in plain English
In ChatGPT, press "Explore" in the top left corner.
Press "Create a GPT" to build your custom generative AI agent.
This will take you to the page where you can enter the instructions that your GPT should execute under the "Create" tab using text prompts. Just have a conversation as if you were asking ChatGPT, including the back and forth that you would have with any expert coder.
Your GPT application will be built in the background.
Under the "Configure" tab you can edit and possibly change what ChatGPT understands your need is.
And in the preview window on the right, you can see and test a preview of your personal GPT, including an app icon that DALL-E 3 automatically generates from your instructions.
Recommended by LinkedIn
Use cases for the GPT
My first GPT
And all this without having written a single line of code. Simply by wording my requirements in plain Englisch.
There are plenty of other use cases, because practically everything that can be done with a keyboard should be automatable with a multimodal GPT:
GPTs can
Limitations
As of November 2023, GPTs are only available as a beta. While the functionality delivers impressive results, the connection between the downstream services regularly fails.
This means that GPTs are not yet suitable for productive systems.
However, you should acquire an initial feel for the simplicity of the system and the wealth of possible applications immediately. For example, GPTs partially replace applications in Jupyter notebooks (link to my post on how to use Jupyter Notebooks in Anaconda framework).
This post first appeared on my blog webmemo.ch/ai/gpt-store-guide-build-custom-ai
IT-driven Innovations
1yAI generated output is sometimes difficult when it demonstrates our flaws, short comings and humanity of our data our personal AI is based upon. We can't run away from it because it's there. It's hard to show people our own inadequacies but it makes us who we really are as producers, makers, knowledge workers, futurists, artist and the like. Once we spend time honing our own AI, we might not produce mutch of relevance for others but might become a better human being for it. AI humbles us.