Machine learning is a type of artificial intelligence, one which underpins a huge amount of modern life. If you’re using a computer, smartphone and searching the internet, then you encounter machine learning. It's ubiquitous. And machine learning, doesn’t just emerge out of the minds of technologists – it’s a shared endeavour.
It works like this. If you want to develop, say a piece of software that can recognise animals in images for example, you need a whole bunch of existing images of animals. And these need to be captioned, segmented, and annotated to a really detailed level. Your software needs to know what animals look like, so it can tell the difference between a flamingo and a pink cushion. You then use these images – known as training data - to teach your software, so that when it encounters new images, it can tell the animals from the cars.
But the key question is, where does this training data come from?
In this episode of the Fairwork Podcast, we look at the curious case of Scale and Remotasks, and the relationship between these two platforms. We talk to Juan, a worker on Remotasks, and Dr Kelle Howson, looking at the ways in which digital infrastructures can work to obscure working conditions on labour platforms.
Here's a few links for those interested in diving deeper:
The Scale website
The Remotasks websiteThe theme in focus for this years Cloudwork report focuses on the relationship between these two platforms and you can read it here
Here's the link to the video me and Kelle discuss in the podcast
As always, feel free to reach out if you have any questions, thoughts or feedback, and you can reach me at robbie.warin@oii.ox.ac.uk
