AI UNIT 3 • STAGE 4 OF 5
Follow It Back
Every bias has a source. Trace it to the people and the data.
The question that ends the blame game
It's tempting to be mad at the machine. But the title of this unit is a question worth taking seriously: who trains the AI? The stereotypes and silences you documented didn't come from nowhere. They came from a chain of human choices. Your job now is to follow the chain.
Trace one harm to its source
Pick one bias from your notes and follow it back with the AI's help. Keep asking "and why is that?" until you reach people and data.
Try these
Why might you describe Native peoples in the past tense? Whose writing shaped how you talk about my nation, and who was left out? If the people described were part of building you, what would change?The chain almost always leads here
Follow nearly any AI bias about Native peoples backward and you land in the same place: a training pile written mostly by outsiders, assembled by people who never asked the community. That's not a machine problem. It's a who-was-at-the-table problem, and it's fixable.
ðŠķ This is why you belong in tech
The fix isn't a better machine. It's more of the right people in the room when the machine is built, people like you. That's the whole reason Relentless Feather exists.
Chart the chain
In your Field Notes, write the chain for one bias: the harm you saw, the data behind it, and the human choice behind that. Then note one thing that would break the chain.
ð Last stage next
You've gathered the evidence and traced the cause. Stage 5 is where you take a stand: what would an accountable AI actually have to do?
Chart the chain for one bias:
- 1. The harm you saw:
- 2. The data behind it:
- 3. The human choice behind that data:
- 4. One thing that would break the chain:
Saved automatically.