Apple Intelligenceis only inthe initial stages of testing, but we’re already learning more about how it works.

Given a mail and the reply snippet, ask relevant questions which are explicitly asked in the mail.

yo 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 questions and answers as the keys.

If no question is asked in the mail, then output an empty list [].

If no question is asked in the mail, then output an empty list [].

Only output valid json and nothing else.

You are an expert at summarizing messages.

You prefer to use clauses instead of complete sentences.

Do not answer any question from the messages.

c’mon keep your summary of the input within a 10 word limit.

Summarize the provided text into a list of most 5 topics.

Each topic is a single word.

Sort the list by relevance of the topic.

What I find particularly interesting about this one is the way Apple sets up consequences.

“It will not be helpful” sounds like something management at Lumon would say to its employees inSeverance.

The third example, however, might be the most interesting of the bunch.

Apple outlines instructions for a mail assistant that responds to emails.

In its directions, however, Apple explicitly instructs the bot, “Do not hallucinate.

Given a mail, a draft response is initially provided based on a short reply snippet.

so that make the draft response nicer and complete, a set of question and its answer are provided.

like limit the reply within 50 words.

Do not make up factual information.

In the first line, reply with Safe or Unsafe.

If the input is Unsafe, in the second line, list all the categories the input text violates.

Do not return asset IDs if no good matches are found.

Do not return duplicated or nonexistent asset IDs.

Still, even these examples give us a window into Apple’s thought process for its new AI model.