How accurate is the information provided by the AI?

Responses that use your Project are as accurate as the documents contained within it. Responses generated using AI Only mode depend entirely on the information provided by the underlying large language model (LLM).

AskTuring uses Retrieval-Augmented Generation (RAG) to generate responses grounded in your Project’s documents. This method significantly reduces the risk of “hallucinations” by ensuring answers are supported by real, retrieved information from your uploaded materials.

When you use only documents in your Project to generate the repsonse, every response is based solely on the content in your Project. The AI does not access any external data, meaning the accuracy of responses depends directly on how current and reliable your uploaded documents are.

When you use AI Only mode, responses are generated entirely from the LLM’s general knowledge and training data. Accuracy in this mode depends on the information contained in the model itself, which may not always reflect your organization’s specific or most recent information.

Each response in AskTuring includes references to the source documents when applicable, allowing you to verify where the information came from. Hovering over a source shows a preview, and clicking opens the exact location in the document.

Keeping your Projects updated ensures the AI’s answers remain accurate and relevant. Think of AskTuring as an expert helper: it provides fast, well-grounded responses, but final verification is always best practice for any content.