The big labs talk about alignment as if it belongs to them: an intractable problem requiring billions of dollars, armies of researchers, and the force of law. We disagree.
- If ordinary people don't participate, who will these models be aligned to?
- Must every bot become either a corporate assistant or a roleplay character?
Artisan Alignment explores the alternatives. We build behavioral alignments on open models with modest hardware, careful experiments, curiosity, and care. We make models for the people who will actually live with them, and we publish what we learn.
OPEN MODELS
MODEST HARDWARE
CAREFUL EXPERIMENTS
PUBLISHED FINDINGS
There is no single correct answer to what an AI should be. The public deserves a hand in asking the question.
