#openai assistants restrict access to internet

1 messages · Page 1 of 1 (latest)

tender island
#

I want to make an openai assistant strictly for IT requests. I don't want users to ask random questions to it and I want it to answer purely based off what data it has been given for retrieval. How can I do this?

hardy ospreyBOT
#

Hi @tender island. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

OpenAI's GPT models, including the AI assistants, are trained on a diverse range of internet text. However, they do not have the ability to access or retrieve personal data unless it has been shared with the model in the course of the conversation. They are designed to respect user privacy and confidentiality.

To create an AI assistant strictly for IT requests, you would need to carefully design the prompts and possibly use a fine-tuned model. Fine-tuning is a process that involves additional training on a specific dataset, which could be a dataset of IT-related conversations in your case.

However, the AI models do not have the ability to access the internet or any external databases for information retrieval unless explicitly programmed to do so in combination with other services. The responses they generate are based on the patterns they learned during training.

If you want to restrict the assistant to answer purely based on given data, you might want to look into Azure's OpenAI service, which allows you to connect the model with your data, giving it the ability to retrieve and utilize data in a way that enhances the model's output. The data is retrieved from designated data sources based on the user input and provided conversation history. The data is then augmented and resubmitted as a prompt to the model, giving the model contextual information it can use to generate a response.

Please note that while the models are designed to avoid generating inappropriate or unsafe content, they might not be perfect and may sometimes need guidance to provide the desired output. You can use the system level instructions to guide the model's behavior throughout the conversation.

#

@tender island kapa.ai is still learning and improving, please let us know how it did by reacting below

tender island
#

@hardy osprey can I use a fine tuned model with an openai assistant?

hardy ospreyBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Based on the information from the OpenAI community forum, the Assistants API does not currently support the use of fine-tuned models. This means that you cannot directly integrate a fine-tuned model into the Assistant API.

However, you can use fine-tuned models separately through the OpenAI API. Fine-tuning allows you to customize a model for your specific application, providing higher quality results, the ability to train on more examples, token savings due to shorter prompts, and lower latency requests.

Fine-tuning is currently available for certain models including gpt-3.5-turbo, babbage-002, davinci-002, and others. Once a model has been fine-tuned, you won't need to provide as many examples in the prompt, which can save costs and enable lower-latency requests.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@tender island kapa.ai is still learning and improving, please let us know how it did by reacting below

tender island
#

@hardy osprey what's the limit of functions in an openai assistant

hardy ospreyBOT
#

Let me check the knowledge base for you. One moment! :books:

#

The OpenAI Assistant API allows you to call functions intelligently based on user inputs. However, to prevent infinite or lengthy loops, there is a limit on the maximum number of function calls. In the example provided in the OpenAI Cookbook, this limit is set to 5 function calls. This is not a hard limit set by OpenAI, but rather an example of how you might want to manage function calls in your own implementation.

In terms of parallel function calling, newer models like gpt-4-1106-preview or gpt-3.5-turbo-1106 can call multiple functions in one turn. However, the exact limit on the number of functions that can be called in parallel is not specified in the provided sources.

Please note that the chat completions API does not execute the function; instead, it generates the JSON that you can use to call the function in your own code.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@tender island kapa.ai is still learning and improving, please let us know how it did by reacting below

tender island
#

@hardy osprey can I tell the AI to answer only IT related questions in the system prompt instructions? Would that work and decline messages not regarding IT?

hardy ospreyBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Yes, you can guide the AI's behavior by providing a system message at the start of the conversation. The system message is used to set the behavior of the AI model throughout the conversation. For example, you can specify that the AI is an IT assistant and its purpose is to answer IT-related questions.

Here's an example of how you might set up the system message:

'messages': [
    {'role': 'system', 'content': 'You are an IT assistant. Your purpose is to answer IT-related questions.'},
    {'role': 'user', 'content': 'What is my question?'}
]

However, please note that while the system message can guide the AI's behavior, it may not perfectly filter out all non-IT related questions. The AI does not have the ability to decline messages, but it can be guided to stay on topic. If a user asks a non-IT related question, the AI might still respond, but you can instruct it to steer the conversation back to IT-related topics.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@tender island kapa.ai is still learning and improving, please let us know how it did by reacting below

tender island
#

@hardy osprey how do I perfectly filter it

hardy ospreyBOT
#

Let me check the knowledge base for you. One moment! :books:

#

While you can guide the AI's behavior using system messages, there's no guaranteed way to perfectly filter out all non-IT related questions using the AI alone. The AI models are designed to generate responses based on the input they receive, and while they can be guided to stay on topic, they don't have the ability to decline or filter out messages.

However, you can add a moderation layer to the outputs of the Chat API to prevent content that violates certain rules from being shown. This could be used to filter out responses to non-IT related questions. OpenAI provides a moderation guide that you can follow to implement this.

Additionally, you could implement a pre-processing step in your application to check if the user's message is IT-related before sending it to the AI. This could involve keyword checking, intent classification, or other natural language processing techniques. If the message is not IT-related, your application could respond with a default message indicating that only IT-related questions are accepted.

Please note that these solutions would require additional development work on your part and are not built-in features of the OpenAI API.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

#

@tender island kapa.ai is still learning and improving, please let us know how it did by reacting below

tender island
#

@hardy osprey can we use moderation with assistants API?

hardy ospreyBOT
#

Let me check the knowledge base for you. One moment! :books:

#

Based on the provided knowledge sources, there is no direct mention of using moderation with the Assistants API. The moderation guide provided by OpenAI specifically refers to the use of moderation with the Chat API.

However, the principles of moderation could potentially be applied to the outputs of the Assistants API in a similar manner. This would involve adding a moderation layer to the outputs of the Assistants API to prevent content that violates certain rules from being shown.

Please note that implementing moderation with the Assistants API would likely require additional development work on your part and is not a built-in feature of the OpenAI API.

For the most accurate and up-to-date information, I recommend checking the official OpenAI documentation or reaching out to OpenAI support.

Sources:

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: