Woudl be nice if we could choose who reviews our noms

To be clear, i dont mean we get to pick our friends. What i mean is, we couls choose of a nomination gets checked by emily or the community. For example

I submitted metal doves on a wall, ghe commjnity would have 100% accepted them but emily rejected it, so for that one ill now need to waste an appeal to get it acceoted, but had i been allowed to select community, i would have had to wait nllinger but not waste am appeal

Ive been submitting trail markers that also double as info sogns, emily loves them has accepted every one thats gone to her, butthe community hate them because they arent completely locked in place (they are held in place by metal … wires? Dunno what the actual name is) so although they had been there for half a decade and have a wrbsite literallt showing where each one is, they get rejected for tenporary, so again, im hiving tk waste appeals to get them accepted, whereas had they gone to emily they would have been quickly accepted

1 Like

That would mess up the training of the ML. From Niantic’s perspective, it would be extra coding to create something that would denigrate their system.

Instead, I suggest rewording, using a different picture if possible, and trying again. Sometimes it’s random what goes to AI and what doesn’t - but sometimes it’s based on a keyword or something it sees in a picture.

I believe Emily get them all first.

1 Like

I thought emily selected specific things straight away, not everything.

Who knows :woman_shrugging:t2:

The ML (Machine Learning) part of maintaining AI is to forever keep feeding it data created by humans, for calibration. The human training for Google AI recommends humans deciding 10%, randomly chosen, if feasible.

Secondly…
When AI looks at a nomination, it returns yes/no, and a percent confidence. The company (in this case, Niantic) decides what to do with what percents. For example, AI might say it’s 65% sure this should be rejected. Niantic might say: we need 85% confidence to reject. So in this example, it would roll to humans to review. (Not the same as the randomly chosen set.)

2 Likes

Found where Niantic mentioned the reason I listed first:

We also continuously audit our models by holding out a percentage of predictions to be reviewed by humans so that we have a fresh assessment of live model performance.