The Wheelhouse · Digitization and AI

The Newest Machine

Digitization and AI can widen memory’s reach—or extract it. Community authority must steer.

Overview

Inside the collection

Digitization can help families and communities preserve photographs, home movies, oral histories, and documents while connecting separated records. NMAAHC programs model family-history support, community curation, fellowships, and digital-first preservation designed around community participation and control.

Artificial intelligence raises a sharper governance problem. The Algorithmic Justice League emphasizes voice, choice, transparency, and agency in confronting AI harms; NIST found demographic differentials in most face-recognition algorithms it evaluated in 2019. FishyGrits therefore treats AI as a contested instrument: useful only under questions of consent, refusal, training data, accountability, ownership, and benefit.

Three close readings

Archive entries

Documented history

Community Curation, Not Data Extraction

Preservation should increase a community’s control over its own record.

Read Community Curation, Not Data Extraction

Documented history

What the Face-Recognition Test Found

Measured error can expose unequal technical consequences.

Read What the Face-Recognition Test Found

Editorial interpretation

Who Holds the Wheel?

The central AI question is governance, not admission.

Read Who Holds the Wheel?

Research notes

Source trail

These sources inform the archive’s account; citation does not imply an institution’s endorsement of FishyGrits.

NMAAHC — Center for Digitization and Curation (opens in a new tab) — Family history, community curation, home movies, and fellowships.
NMAAHC — Community Curation Program (opens in a new tab) — Digital-first preservation with community control.
Algorithmic Justice League — About (opens in a new tab) — Accountability principles centered on voice, choice, transparency, and agency.
NIST — 2019 face-recognition evaluation (opens in a new tab) — Evidence of demographic differentials and application-dependent effects.