Noah Turk Ratings
Workers feel this requester pays well
Noah Turk Wage History
Top Worker Reviews
Beat the System! (Creation) - $1.10 +0.01 bonus
$7.50 / hour
00:08:53 / completion time
ProsThese were pretty good money the last time around and they aren't that difficult once you figure out the blind spots.
EDIT: I almost timed out on one of them after submitting "Beating the System" twice because it takes forever to get back to you on the ones that are supposed to be neither correct nor incorrect. This is a flawed system.
ConsThey are giving the bot way too much time to figure out the statement now. Even if you nail it on the first try you could be waiting for 1:30 for the bot to admit that it's wrong. I don't remember what the bonuses on these were like so I'm not sure if this is going to be worthwhile or not.
Advice to RequesterPlease reduce the time for the model to make it's decision on the neither correct/incorrect requests.
Fool the model - Visual Question Answering - $0.20 +0.26 bonus Confirmed!
$40.39 / hour
00:00:41 / completion time
ProsIf you have a good intuition for writing the kinds of questions that fool the AI pretty consistently, these can go by really fast. The bonuses are a nice plus, but they take forever to pay out and there's no real way to meaningfully track them.
ConsYour mileage will always vary wildly on these, if you can even get them to load at all. When it's not the Heroku "build something" message, it can take up to minutes for the picture to even appear. Some are way more difficult than others to come up with unique questions for, let alone clever questions that fool the AI.
Advice to RequesterDon't post more HITs than there are photos to work on. It's very discouraging to constantly get faced with a "build something" screen instead of a real HIT.
Write sentences and fool the AI - $1.00
$7.50 / hour
00:08:00 / completion time
ProsThe AI has gotten dumber since the last time I did these and seems much easier to trick.
The last three before you get capped go quicker than the first because the onboarding process is out of the way.
ConsThere are extra fields in the HIT now where you have to explain
1) Why you believe your statement to be be correct, incorrect, or neither correct or incorrect
2) Why you think the AI will be stumped by the statement.
Cap is 4 HIT's then you aren't allowed to do anymore.
There's an onboarding process for the first HIT that eats up time.