You open something you wrote three years ago and think, Who gave this person so much confidence? Unfortunately, the author is you.
Now imagine an archive with hundreds or thousands of such snapshots. Could you ask an AI, “What did I think about changing jobs back then?” Yes, if it searches the archive and quotes the relevant posts. No, if you expect a copy of your mind.
Keep the reasons, not just the verdicts
“I wanted a different job” is a conclusion. “I wanted more control over my evenings” explains the decision. Dates, context and the reasoning behind an opinion make old writing useful rather than merely nostalgic.
Researchers describe how people use external tools, such as notes and reminders, to reduce mental effort. That does not show that a blog stores consciousness. It does explain why searchable records can help with remembering.
Search first; ask the AI to answer second
A practical workflow is simple: ask a question, find relevant posts, give the excerpts to the AI, then insist on a source-backed answer. This is the idea behind systems that retrieve documents before producing a response.
A good answer might say: “In a post from one year, you favored option A because of time. In a later post, you changed your mind after learning B.” A poor answer sounds like a confident therapist who has never actually read the diary.
Less Horcrux, more library card
In Harry Potter, Tom Riddle's diary is a terrifying piece of the past that talks back. An archive may feel a little like that when AI brings back old words. But it isn't a Horcrux. No soul is preserved, and the model can't predict every future choice.
The point is almost the opposite: you should be able to disagree with your past self. Keep the original post, mark what's outdated and explain why your view changed.
The catch: an archive can preserve mistakes too
Old assumptions, jokes without context and private details do not become reliable just because an AI can find them. Keep dates and sources, check time-sensitive claims against newer evidence, and avoid putting identifying information in public records.
A thousand posts won't make you immortal. They might give your future self a remarkably good filing cabinet—with a search box and, occasionally, an embarrassing receipt.
References (3)
- Risko & Gilbert (2016), Cognitive Offloading: https://doi.org/10.1016/j.tics.2016.07.002
- Clark & Chalmers (1998), The Extended Mind: https://doi.org/10.1093/analys/58.1.7
- Lewis et al. (2020), Retrieval-Augmented Generation: https://arxiv.org/abs/2005.11401
