OpenAI announced new changes to its file search system last week, allowing developers more control when asking AI chatbots to choose responses. The improvement has been made to ChatGPT ’s API and will allow developers to not only inspect the behavior of the chatbot’s response retrieval method, but also to fine-tune its behavior. This way, developers can ensure that only the desired responses are chosen
A previous report claimed that the company is planning to launch another AI model called "Strawberry" that can improve math and logic in ChatGPT
The AI company announced the changes to the API in a post on X (formerly known as Twitter), and basically, the upgrade improves the file search controls in the Assistant API. It allows developers to check the results the chatbot has selected and make further adjustments according to their requirements
APIs are different from consumer-focused ChatGPT apps and websites, and while the interface that end users see is fine-tuned by OpenAI and tuned to behave in a certain way, developers who are either building internal tools for companies or integrating the chatbot into different apps and software need more freedom
This may be because while the public version of ChatGPT is built for general purposes, the API version is used for one specific function, and to excel at it, users require the AI to make no mistakes, and return higher-quality responses
Until now, developers didn't have the option to adjust the API to make the chatbot generate responses relevant to specific use cases, however, with the new control options, this will change, and OpenAI, on its support page, highlighted how this will work
Developers can now examine file search responses, and the file search tool in the Helper API picks answers that it thinks are relevant to a given query. However, developers will now be able to check the responses the AI picks and test the information it has generated in previous runs. This information is said to help them provide more insight into the tool's operation
Furthermore, developers can modify the settings of the result classifier that is used to search for information to generate responses. By choosing a classification between 0.0 and 1.0, they can control which information the AI selects and which it ignores