THE KALEIDOSCOPE PROJECT

Shifting Big Tech's Top-Down Push

To counterbalance the top-down push that constrains Public Interest AI from advancing Big Tech's interests, we can turn to Community Contextual Data, which encompasses public sentiment and data on deployment context.

Currently, AI has a terrible reputation (only 9% of Americans believe AI will do more good than harm) and questionable performance in deployment contexts (95% of AI solutions fail when moving from pilot to production).

Community Contextual Data is critical both to the general goal of Public Interest AI and the interests of the individual groups within it. When it comes to the power of actually capturing this Community Contextual Data, the current top-down power structure flips so that communities themselves have the leverage.

Explore the power structure behind Community Contextual Data below and how we believe community ownership over it is central to our Operating Hypothesis.

Power Behind Community Contextual Data

Communities

1 of 6

Where it starts

  • The source of the data
  • Define what they want from AI themselves
  • Hold the first decision

What they gain

  • Agency over how or whether they interact with AI
  • Benefit that routes back to them, with contributors compensated
  • Solutions that answer needs they named

What they contribute

  • Lived experience
  • Public sentiment, including where people disagree
  • Ground truth about what a deployment does

What shifts

  • From absorbing most of the harms to setting the need
  • From ethical choices made for them to choices made by them

Think that this work could be improved? We would love your feedback!