#48 Ghosts in the machine

This week, Arj and Jordan look at the human labour powering many AI solutions - solutions that we're encouraged to view as magic in a box. The way this labour is being used - much of it through sub-contracting arrangements in the developing world - raises worker welfare and privacy issues, among other concerns.

Arj and Jordan explore this issue through a couple of news stories - one that reveals the workers involved in labelling images taken by some Roomba robot vacuum cleaner models, and a second story that reveals the role played by low-paid Kenyan workers in training ChatGPT to not be violent, racist and sexist.

Links:

Article about the ChatGPT making jobs redundant (NYPost) https://www.news.com.au/technology/innovation/inventions/the-jobs-ai-could-be-about-to-take-over-forever/news-story/3969af5aed12008263710e264876d5d4

Article about Roomba recording sensitive images (MIT Technology Review) https://www.technologyreview.com/2022/12/19/1065306/roomba-irobot-robot-vacuums-artificial-intelligence-training-data-privacy/

Article about OpenAI using Kenyan workers to train ChatGPT (TIME) https://time.com/6247678/openai-chatgpt-kenya-workers/

Article in NOEMA magazine https://www.noemamag.com/the-exploited-labor-behind-artificial-intelligence/

Book - The Costs of Connection (goodreads review) https://www.goodreads.com/en/book/show/43506400

Credits:

Music by Bensound.com