The personal score of a message indicates how much of it was generated resorting to your personal data. For example, a personal score of 65% shows that Personal AI used 65% of content from your memory stack when generating that answer.
There are two factors that could lead to a high personal score response:
Since the personal score improves if the content of the memory blocks is very similar to the content of the question or prompt, it makes sense to stack new memories in question/answer pairs. This is especially useful for questions that have always the same answer, where there’s usually no nuance.
Instead of stacking the following sentence…
… you can stack it as a question/answer pair:
This method works well when messaging your AI, but for longer documents it may become time-consuming to write all the questions and answers. But, since this is such a gamechanger for personal scores, Personal AI offers a QnA rewrite feature to automatically format documents with this structure. Just open the document you want to reformat and click the Generate QnA button. Learn more about QnA rewrite and other training tools.
Reinforcing memories in Personal AI means uploading more data that supports a specific view or topic. When generating a new response, your AI will have more experience guiding it to provide an answer. This strategy is better for answers that require a more nuance when compared with QnA.
First, identify which answers you want to improve. Then, gather as many documents that contain data relevant to the topics connected to that answer. The more information you upload and stack, the more the personal score will increase over time.