Prolific Academic Survey Alerts

MollyR

I am working on the assumption that these studies are generated and controlled by AI, which has not been programmed to respond to user messages.

elsien

The two about the fault on the study I didn’t do on prolific. I used their feedback section on the study itself.

The other one where you’re following the progress of the bot through a search now has the edit button and instructions  but despite having done the tutorial I still can’t make nor tail of how it’s meant to work. I strongly suspect their instructions simply don’t make sense. 

And I’m on a mission to stop them from automatically accepting cookies.  Or presuming  that everyone lives in the states 😀

ele_91

MollyR said:

A couple of days ago I did one of their surveys where you have to make judgements on AI-manipulated images.  There is an immensely long "training" session, where if you do not give the exact scores they like to questions on an image you are told what you should have said, and told to do the questions again.  Some of their own "answers" are very clearly wrong, but without parroting these scores you do not get to progress.  I spent an hour on this without reaching the end of the multiplicity of questions.  It is one of those studies where you have to terminate it after an hour, and I did so;  later I received a message telling me to return it, on the grounds that "Tasks ran out while connecting to the study."  (It was still on offer, with many places remaining.)  I have refused to return it, and raised a ticket with Prolific, because in the case of a similar study a couple of weeks ago I discovered it had been marked as Returned without my knowledge, and I am fairly sure they will do the same again.  I suspect that all of these studies are not only about, but run by, AI bots, because you can never get an answer from them, and they come up with "decisions" which are grossly unjust.  Calling themselves "Humans" is rather a giveaway!   If I give them an hour of my time which I will never get back again, I want to be fairly paid for it.

I have had replies to messages from this researcher before and they have messaged me so I don’t think they’re bots. I think they probably run so many studies they just don’t respond. 

onemouseplace

elsien said:

The two about the fault on the study I didn’t do on prolific. I used their feedback section on the study itself.

The other one where you’re following the progress of the bot through a search now has the edit button and instructions  but despite having done the tutorial I still can’t make nor tail of how it’s meant to work. I strongly suspect their instructions simply don’t make sense. 

And I’m on a mission to stop them from automatically accepting cookies.  Or presuming  that everyone lives in the states 😀

I only did the tutorial and about 3 examples for that one, and haven't been given any more so I must have done something wrong!  Shame, as once I got into the swing of it, I was quite enjoying it (and it was very similar to other work I've done).

snowwhyte

I never got any tasks after doing that tutorial so I'm fairly sure I failed it. I am lucky that I'm on a slightly different AI task that currently has a lot of tasks and that I enjoy more than the original fact checking study. 

lou-rose

Now we seem to be getting paid in $, is everyone just using the paypal exchange rate to convert to £ (which takes off about 4%), or has anyone found a better way?

snowwhyte

I've been using the PayPal exchange and the pound is currently down against the dollar so the studies are paying slightly more than they used to for now. Theoretically, if you use a Revolut card you could withdraw dollars from PayPal and then convert them within the app but I'm not sure if their exchange rate is much better than PayPal's rate.

onemouseplace

snowwhyte said:

I never got any tasks after doing that tutorial so I'm fairly sure I failed it. I am lucky that I'm on a slightly different AI task that currently has a lot of tasks and that I enjoy more than the original fact checking study. 

Yes, I'm doing that one at the moment as well. I like it, other than it's a pain having to scroll up and down some of the massive chunks of prompt text (I've resorted to copy and pasting it to a separate doc so I can search more easily).

snowwhyte

onemouseplace said:

snowwhyte said:

I never got any tasks after doing that tutorial so I'm fairly sure I failed it. I am lucky that I'm on a slightly different AI task that currently has a lot of tasks and that I enjoy more than the original fact checking study. 

Yes, I'm doing that one at the moment as well. I like it, other than it's a pain having to scroll up and down some of the massive chunks of prompt text (I've resorted to copy and pasting it to a separate doc so I can search more easily).

Are you using control F? I normally pick the most relevant word from the claim, do a control F search and then tab down until I find the relevant, supporting statement. It doesn't always work but it can speed things up. 

onemouseplace

snowwhyte said:

onemouseplace said:

snowwhyte said:

I never got any tasks after doing that tutorial so I'm fairly sure I failed it. I am lucky that I'm on a slightly different AI task that currently has a lot of tasks and that I enjoy more than the original fact checking study. 

Yes, I'm doing that one at the moment as well. I like it, other than it's a pain having to scroll up and down some of the massive chunks of prompt text (I've resorted to copy and pasting it to a separate doc so I can search more easily).

Are you using control F? I normally pick the most relevant word from the claim, do a control F search and then tab down until I find the relevant, supporting statement. It doesn't always work but it can speed things up. 

Yes I am (although you are right - it doesn't always work!) - I just find it easier to have the document in a separate tab so I can compare it with the exact wording of the claim more easily.