It feels like a missed opportunity that ChatGPT doesn't know much about itself and related OpenAI products, e.g. meta knowledge like "which version of ChatGPT am I talking to?", "what does the o in GPT-4o stand for?", "can I use GPT APIs with a Plus subscription?" or other similar questions.
Is there a published reason for this or is that something I should raise elsewhere?
#Missed Opportunity? ChatGPT knowledge about itself and other OpenAI products seems limited
magic dagger
wooden trenchBOT
Hi @magic dagger. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
ChatGPT and other OpenAI products are designed with specific capabilities and limitations. For instance, ChatGPT is a powerful generative AI model that can answer questions and complete tasks using natural language. It can also use tools like web browsing and image processing to enhance its capabilities. However, it has a knowledge cutoff due to its static training data, which means it may not have the most current or specialized information, including meta knowledge about itself or other OpenAI products source (https://help.openai.com/en/articles/9260256-chatgpt-capabilities-overview#h_87b98a8413).
As for your specific questions:
- "Which version of ChatGPT am I talking to?" - This information is not inherently known to ChatGPT.
- "What does the 'o' in GPT-4o stand for?" - This information is not provided in the knowledge sources.
- "Can I use GPT APIs with a Plus subscription?" - The ChatGPT API and ChatGPT Plus subscription are billed separately. The API has its own pricing, and the Plus subscription covers usage on chatgpt.com only and costs $20/month source (https://help.openai.com/en/articles/7039783-how-can-i-access-the-chatgpt-api).
If you believe that ChatGPT should have more meta knowledge about itself or other OpenAI products, you can provide feedback to OpenAI. If you're using a GPT developed by the community and the developer has opted in to share a contact address, you can share feedback directly. If you're encountering issues with the OpenAI platform itself, you can open a new chat from the Help Center source (https://help.openai.com/en/articles/8798889-how-can-i-use-gpts).
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
magic dagger
@wooden trench it didn't know to send me here, either. Also the answer to your third point was not what I asked -- I'm more interested in why was the particular choice made to not make knowledge about itself available to ChatGPT. Is there a published reason? Is there a technical limitation? Would it make sense to have a specialized GPT instance that the general one knows to point to?
wooden trenchBOT
Let me check the knowledge base for you. One moment! :books:
The decision to limit ChatGPT's knowledge about itself and other OpenAI products is likely due to a combination of design choices and technical constraints.
One reason could be related to privacy and safety. OpenAI has implemented systems to ensure that GPTs adhere to usage policies, which includes preventing harmful content and impersonation. This could potentially extend to limiting the amount of meta knowledge the model has about itself or other OpenAI products to prevent misuse source (https://help.openai.com/en/articles/8554407-gpts-faq#h_56e44d6ee1).
Another reason could be the static nature of the model's training data. The language models powering ChatGPT are trained on a significant amount of the world’s information, but they have a knowledge cutoff because their training data is static. This means that the model doesn't have the ability to learn new information after it has been trained, including updates about itself or other OpenAI products source (https://help.openai.com/en/articles/9260256-chatgpt-capabilities-overview#h_87b98a8413).
As for the idea of a specialized GPT instance that the general one knows to point to, it's an interesting concept. However, it's not clear from the provided knowledge sources whether this is technically feasible or if it aligns with OpenAI's design principles.
If you have further questions or suggestions, I recommend reaching out to OpenAI directly for more information. They may be able to provide more insight into their design choices and future plans.
Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:
magic dagger
@wooden trench You mentioned providing feedback to OpenAI directly, can you point me at a good way to do that? Is there a specific channel in this discord server that I could use or some other method that is preferred by OpenAI? Thanks!
wooden trenchBOT
Let me check the knowledge base for you. One moment! :books:
OpenAI has several channels through which you can provide feedback:
-
Developer Forum: You can share your feedback on the OpenAI Developer Forum (https://community.openai.com/). This is a platform where developers and users of OpenAI products can discuss, share ideas, and provide feedback source (https://help.openai.com/en/articles/8550641-assistants-api-v2-faq#h_c5908ad4b4).
-
Model Behavior Feedback Form: If your feedback is related to the behavior of the model, you can use the Model Behavior Feedback Form (https://openai.com/form/model-behavior-feedback). This form is specifically designed to collect examples where the models did not meet user expectations, which helps OpenAI improve their models source (https://community.openai.com/t/736062).
-
OpenAI Discord Server: You can also join the OpenAI Discord Server (https://discord.com/invite/openai) and share your feedback there. They have a Showcase section for interesting projects, and you might find relevant channels for your feedback source (https://community.openai.com/t/623807).
-
Help Center: If you have an account with OpenAI, you can log in and use the "Help" button to start a conversation. If you don't have an account or can't log in, you can still reach OpenAI by selecting the chat bubble icon in the bottom right of help.openai.com (https://help.openai.com/) source (https://help.openai.com/en/articles/6614161-how-can-i-contact-support).