Just had a doctoral supervision and the researcher had the most appalling experience with pseudo-participants - AI pseudo participants. Turns out these people now use LLMs to complete forms to be included in research, and then try to use them to complete interviews and get the money! - ThreadSky

urbaneprofessor.bsky.social • 29 days ago

Just had a doctoral supervision and the researcher had the most appalling experience with pseudo-participants - AI pseudo participants. Turns out these people now use LLMs to complete forms to be included in research, and then try to use them to complete interviews and get the money!

Comments

magnets.bsky.social•29 days ago

Bloody hell!

urbaneprofessor.bsky.social•29 days ago

I know!

hbsj.bsky.social•29 days ago

It looks like you're trying to undertake rigorous research ...

snoopdougiedoig.bsky.social•29 days ago

Blimey!
It's all getting a bit dystopian.

erinjlux.bsky.social•29 days ago

we've had a real problem with this in our work as well - happy to chat about it if you're interested!

urbaneprofessor.bsky.social•29 days ago

I might put the researcher in contact with you, if you want?

erinjlux.bsky.social•29 days ago

Sure! Not sure I'll be much use in terms of practical ways to avoid it in future but I think knowledge-sharing about how common this is might be helpful if we want to keep doing this kind of engagement work

lmicharlie.bsky.social•29 days ago

oh no, I've decided to unread this and pretend it's not happening

urbaneprofessor.bsky.social•29 days ago

does make you wonder what the likes of YouGov are doing to keep their online panels legit

lmicharlie.bsky.social•29 days ago

doesn't it just? good luck to anyone conducting research using paid online panels.

kittywashere.bsky.social•29 days ago

YouGov is unpaid, so the incentive isn’t there, but I’m sure those bots once unleashed behave indiscriminately.

1000kindsofrain.bsky.social•29 days ago

We are led by focus-groups. So there's a political incentive to bias the polls in favour of, for example, Farage. It allows someone who is rich to make unpopular policies seem popular, and make popular policies seem unpopular.

kittywashere.bsky.social•29 days ago

The incentive for the individuals using bots to collect rewards for survey completion isn’t there. (These are tiny rewards for enormous amounts of surveys, generally.)

lmicharlie.bsky.social•29 days ago

Ah, I had thought they offered incentives.

kittywashere.bsky.social•29 days ago

Last I knew, they were just incessantly emailing volunteers.

profellenstu.bsky.social•29 days ago

Quite a lot of chat about it in health research already: https://ebm.bmj.com/content/30/3/173
I have to say I am now a wee bit more skeptical of research where recruitment and data generation is purely conducted online, which is sad.

urbaneprofessor.bsky.social•29 days ago

yeah, there is quite a bit of chat all round, and we experienced it on our last project. This was the first time I'd heard of it being AI-powered, which is a bit scarier!

Comments

Posting Rules

Reply