Tell HN: Life Working as AI Data Trainer
Why I started: This is the only job I can find since I moved to UK two years ago.
Pros: I’m making some decent money and saved a bit. At least I can work now.
And learning new maths again 10+ years after graduation is actually fun.
Cons: Instability. No work is guaranteed. One week you could work every hour you can work and the next week, nothing.
Bad actors. A lot of companies are just agencies getting projects from big tech and try to squeeze as much as possible. A lot of shady things happen. A project paying with decent pay ended and the same project came back with a quarter pay per hour. A few were practicing “pay-per-accepted-task” schemes, which is absolutely illegal in their country for not paying for hours worked. The funny thing is, they don’t do that in the US, but once it is outsourced, they think they can do anything.
Good companies that treats the worker fairly with respect are few and far, though they exist. As a general rule of thumb, if it is not possible for you to find your team leader for help, it is one big red flag. If your team leader is rude when you ask questions that are essential for your work, that’s another red flag.
How I feel about the future of the industry: As AI grows, so does the need of data trainers. We can’t build AI products for humans without human feedback. Human trainers are there to stay.
The landscape could be very different in several years though. There are lots of grifters in the AI industry now. As AI boom experiences slower growth, hopefully they will be gone.