Can ai reconstruct your life? one man's digital legacy
Imagine a system that could weave your entire life—photos, videos, messages—into a navigable, interactive encyclopedia. Jeremy, a software developer, has essentially built that, and the implications are both fascinating and unsettling. His project, WhoAmI.wiki, isn't just about preserving memories; it's about using artificial intelligence to unlock narratives buried within the digital detritus of our lives.
From grandma’s photos to a personal wikipedia
It began, as many compelling projects do, with a practical problem. A visit to Jeremy’s grandmother unearthed 1,300 old photographs, a visual avalanche threatening to overwhelm any attempt at organization. What started as a manual effort—meticulously adding dates, names, and locations—quickly morphed into something far grander. He realized he was constructing a rudimentary personal encyclopedia, a framework far better suited to MediaWiki, the software underpinning Wikipedia.
But Jeremy didn't stop there. He integrated large language models (LLMs) into the mix, and the results are startling. A simple test involving 625 photos from a 2012 family trip produced a detailed narrative draft solely from the images, their timestamps, and the visual data they contained. The system didn't just identify faces; it began to infer context, relationships, and events.
The project escalated further. Subsequent trips to Mexico City saw Jeremy feeding the system a torrent of data—Google Maps location history, iPhone 12 metadata, Uber ride records, even bank transactions. The AI cross-referenced this information to pinpoint restaurants visited, soccer matches attended, precise travel routes, and even the music playing at specific locations—details Jeremy himself had forgotten. The ability to reconstruct a lived experience with such granular detail is genuinely remarkable.
WhoAmI.wiki operates locally, keeping all data on the user’s device—a deliberate choice to address privacy concerns. Currently, it's processed around 100,000 personal messages and thousands of voice notes accumulated over more than a decade. It’s a digital life rendered in hyperlinked form, a quantified self project on steroids. But the power of this Technology also raises fundamental questions about data security and the potential for misuse.
The project is open source, allowing others to explore and contribute. However, the sheer volume of personal data involved underscores the need for caution. While the promise of digitally preserving and understanding our lives is alluring, the risks—however carefully mitigated—are undeniable.
Jeremy's creation isn't just a technological feat; it’s a reflection of our increasingly data-rich existence, and a stark reminder that the most intimate details of our lives are increasingly vulnerable to algorithmic interpretation. The question isn't just whether we can build these digital legacies, but whether we should.