Where we use artificial intelligence

No machine drafts, rewrites, lengthens, shortens or polishes your answers. What gets printed in your book are your words, exactly as you said or wrote them.

This is not a matter of style. A sentence your mother would never say is spotted instantly, and a book of memories has to still hold up as a record thirty years from now. If we ever offer a written-up version, it will be a different product, under a different name, and it will say on the cover that a machine wrote it.

To transcribe what you say. If you answer by speaking, a speech recognition system turns your audio into text. Its only job is to write down what it hears. We always keep the original recording, so you can check any sentence against it.

To title each answer. Every answer carries a short title taken from your own words: if you talk about the house on Mayor Street, the title is «The house on Mayor Street». It is extracted, never invented. When an answer gives us nothing good, we fall back on the question's generic title, the same one for everybody.

That is all. There is nothing else.

Your answers, your audio and your photographs are not used to train any artificial intelligence model, ours or anybody else's. They are not sold, not licensed and not shared with third parties for that purpose.

When we choose a transcription provider, the condition for hiring them is that they cannot do it either.

No speech recognition system is perfect, and proper names, small villages and family words are exactly what it handles worst.

That is why you can correct any transcription before it is printed, and why we keep your audio: the recording is what counts, not the text.