Is your ChatGPT content actually yours?
Have you ever used ChatGPT or other AI-powered tools? Do you know how the information you provide on these platforms is utilised? We have examined OpenAI's terms of use and translated the content section into simple language for those who aren't tech-savvy. We will look at two ways of generating content to illustrate.
ChatGPT Direct Interface
When we asked ChatGPT how our input is used by OpenAI, we were informed that the input is not used to train the model as it is pre-trained and we are beyond the training phase.
Here is ChatGPT’s response:
“When you interact with ChatGPT, your inputs are sent to the pre-trained model, and the model generates responses based on patterns it has learned during the training phase. However, your specific input and interactions with the model are not used to fine-tune or update the model on the fly.
It's important to note that while your input is not used to train the model, OpenAI does collect and retain user interactions with the API for various purposes, including improving the performance and safety of the model. However, OpenAI has strict data usage and privacy policies to protect user information.”
However, according to OpenAI's terms of service, if you are using ChatGPT directly, you are giving your permission for your content to be used by OpenAI to improve their services and machine learning. If you don't want your content and information to be used by OpenAI, you can opt-out via this form. But keep in mind that this might limit the results you receive from the model.
External Tools via API
Another way of accessing artificial intelligence such as GPT technology is via an API (Application Programming Interface). This is when you are using an external tool (there are zillions at the moment that are based on GPT technology) that interacts with GPT in the background.
Why use an external tool via API instead of just going directly through ChatGPT?
Think of GPT-3 and GPT-4 like the engine that does the heavy lifting. This engine can be put into a few different models of cars that have different features and comforts.
ChatGPT has minimal features – you interact directly with the chatbot without any additional training, selections, or instructions. Your engine (GPT-3) has been put into a Go-Kart.
Now, imagine you have access to a tool that gives you a multitude of dropdown menus to guide your selections, input boxes to gather the required information and other ways of guiding you to a customised outcome. You’ve now dropped that GPT-3 engine into an Audi – your output has improved without too much thought from you. You’ve been guided through the process.
Want to drive a Ferrari? Look at tools that have examples of generators that you can use that have already been trained in the parameters of what the output should look like. You will be amazed at the results, and how it is fit-for-purpose and requires little editing from you.
The other benefit of utilising a tool via the API is OpenAI’s terms of service outlining that any content or information you provide to GPT through the API is not used to develop or improve services. This means that if you are using external providers of tools built on models such as GPT-3, your input will not be used by OpenAI (unlike when interacting directly through ChatGPT).
Summary
Given the terms of use as explained above, it is worth considering whether utilising external tools would be more beneficial compared to directly accessing GPT technology via ChatGPT. It is essential to carefully evaluate the capabilities and reliability of any external tools to ensure they meet your specific needs, including evaluating the individual terms of use/service of each tool as this will vary.
If you are interested in exploring some of our favourite AI tools for educators, download our information sheet!