It’s almost impossible to have any conversation about the state of the tech industry today and the future of jobs and not end up discussing AI. But those “self-learning algorithms” are often human powered. And they’re giving rise to a new under-class of low-paid workers.
The ability to proceed and automatically react to vast swathes of data holds promise in fields as diverse as self-driving cars through to helping doctors make more accurate and faster diagnoses.
A recent article published by Axios calls the people behind many AI endeavours “AI sharecroppers”. These people work mainly for AI companies from the USA, Europe and China. They come from both western and developing countries and paid little.
Their job is to label objects so that algorithms can be trained. Without them, algorithms couldn’t tell the difference between an egg and an elephant.
I’ve discussed the issue of algorithm fairness before but this is a different question. And it’s not a new one. When new technology has appeared, there have always been groups of workers who have either lost their jobs through an inability to adapt (which may or may not be their fault) or because new jobs appear but are paid at low rates.
In the article I referred to, Microsoft Research’s Mary Gray said the question of what people are paid “is a moral question, not just an economic one”. And while that’s right, it becomes a question of the morals of the people creating those algorithms.
Next time you marvel at how an image search algorithm can tell you what something is or find you a picture of a specific thing, or knows the difference between “good” and “bad” remember that somewhere isn the world a lowly paid worker made that decision.