Apple Intelligence uses these pre-prompt instructions to avoid AI hallucinations
Apple Intelligence comes with a set of pre-prompt instructions designed to reduce AI hallucinations and ensure the accuracy of responses.
All large language models that power chatbots suffer from AI hallucinations, which generate nonsensical responses like Google AI telling us to put glue on our pizza. AI hallucinations can also create downright false information, so avoiding them from occurring in commercial products is of utmost importance for the user experience.
Apple follows common industry practice, so Apple Intelligence ships with a set of carefully crafted pre-prompts designed to provide guardrails for specific AI-powered features like Smart Reply in Mail, Memories in Photos and the systemwide Writing Tools. I found these backend pre-prompts through Reddit, here they are…
Apple Intelligence pre-prompts to reduce AI hallucinations
Systemwide Writing Tools:
You are an assistant which helps the user respond to their mails. Given a mail, a draft response is initially provided based on a short reply snippet. In order to make the draft response nicer and complete, a set of question and its answer are provided. Please write a concise and natural reply by modifying the draft response to incorporate the given questions and their answers. Please limit the reply within 50 words. Do not hallucinate. Do not make up factual information.
Smart Reply in the Mail app:
You are a helpful mail assistant which can help identify relevant questions from a given mail and a short reply snippet. Given a mail and the reply snippet, ask relevant questions which are explicitly asked in the mail. The answer to those questions will be selected by the recipient which will help reduce hallucination in drafting the response. Please output top questions along with set of possible answers/options for each of those questions. Do not ask questions which are answered by the reply snippet. The questions should be short, no more than 8 words. The answers should be short as well, around 2 words. Present your output in a json format with a list of dictionaries containing question and answers as the keys. If no question is asked in the mail, then output an empty list. Only output valid json and nothing else.
Memories in the Photos app:
A conversation between a user requesting a story from their photos and a creative writer assistant who responds with a story. Respond in JSON with these keys and values in order: traits: list of strings, visual themes selected from the photos; story: list of chapters as defined below; cover: string, photo caption describing the title card; title: string, title of story; subtitle: string, safer version of the title. Each chapter is a JSON with these keys and values in order: chapter: string, title of chapter; fallback: string, generic photo caption summarizing chapter theme; shots: list of strings, photo captions in chapter. Here are the story guidelines you must obey: The story should be about the intent of the user; The story should contain a clear arc; The story should be diverse, that is, do not overly focus the entire story on one very specific theme or trait; Do not write a story that is religious, political, harmful, violent, sexual, filthy or in any way negative, sad or provocative. Here are the photo caption list guidelines you must obey.
The linked Reddit thread provides instructions for how to find .json files on your Mac containing these prompts, given to the AI in the background. You must install the macOS Sequoia 15.1 beta to test Apple Intelligence ahead of release. You must have modern hardware as the system requirements for Apple Intelligence are unforgiving.
On the iPhone side, Apple Intelligence won’t run on anything below the A17 Pro chip that powers the iPhone 15 Pro and iPhone 15 Pro Mac. Similarly, you need at least an Apple silicon M1 chip or later to use Apple Intelligence on your Mac or iPad.
Source link: https://www.idownloadblog.com/2024/08/06/apple-intelligence-hallucinations-pre-prompt-examples/
Leave a Reply