Report
A vision for personal AIs as unique as their owners
Kevin Kelly argues that personally trained AIs could reflect their owners’ experiences instead of serving everyone alike.
TLDR
In an October 2026 essay, Kevin Kelly predicts that, if Moore’s Law continues, models as capable as today’s leading LLMs could run on personal devices within a few years. Within about a decade, he says, individuals could afford to train one from scratch. He envisions an “exoself” shaped by its owner’s experiences, handling routine tasks so people can pursue what makes them unique. Kelly notes that personal AIs would still raise trust issues.
Combined views
—
3 Sources, first seen ago
