AI development principles

[PLACEHOLDER — your principles go here. Bullet points are fine and I'll shape them. The structure below is a suggestion: a short numbered list, each with a bolded claim and a couple of sentences of explanation.]

1. Start from the problem, not the model. [Placeholder explanation — a sentence or two on why.]

2. Evaluate before you automate. [Placeholder explanation.]

3. Ship the simplest thing that could work. [Placeholder explanation.]

4. Keep humans in the loop where it counts. [Placeholder explanation.]