top of page
Search

How Has ChatGPT Led to Exploitation in Kenya?

An overview of "OpenAI Used Kenyan Workers on Less than $2 per Hour: Exclusive." by Billy Perrigo from Time.


With ChatGPT being a new technological phenomenon, a symbol of our ever-progressing era of AI, many are extremely wrapped up in this new discovery. With its ability to generate text out of seemingly nothing in addition to answering questions and writing poems and songs, ChatGPT is hailed as one of innovation’s finest. However, like many instances of technology these days, there’s a catch. ChatGPT, being accessible to all users, of course, works to censor all hate speech, inappropriate images, and any other sort of toxicity, but all of it must be human-vetted. While many believe that AI and higher-level technologies such as ChatGPT only utilize our world’s greatest minds to create, it is truly on the back of exploited workers, in this case in Kenya, that they actually ride.


The creator of ChatGPT, OpenAI, reached out to Sama, an outsourcing company in Kenya, to help vet texts for toxicity in the context of censorship. These employees, working for less than $2 an hour, were forced to look at horrifying images and text of extreme hate speech, rape, incest, and extreme violence to “label textual descriptions” and feed their findings into a detector that would be coded into ChatGPT. These workers were provided limited therapy and became “mentally scarred by the work”. While Sama ended their contract with ChatGPT early due to the work’s effect on its employees, it left employees with trauma as well as others without jobs, who depended on the $2 an-hour wage.


Human vetting and labor are still needed with these large projects of sourcing data for algorithms and AI use. As innovation progresses, exploitation still exists, creating an even greater disparity between economic classes as well as national hierarchies. Only through an emphasis on data ethics and acknowledgment from large corporations of the exploitation of third-world countries can we truly have innovation and ethical change in our modern technological era.



Perrigo, Billy. “OpenAI Used Kenyan Workers on Less than $2 per Hour: Exclusive.” Time, Time, 18 Jan. 2023, https://time.com/6247678/openai-chatgpt-kenya-workers/.


Cover image by Kasia Bojanowska on Dribbble.




 
 
 

1 Comment


JJ Toshiba
JJ Toshiba
Feb 04, 2023

There is a great deal of effort in improving algorithms to screen toxic content but we are still far away from replacing human vetting. It is not easy to train the computer to learn nuance, satire, parody, etc. Human judgment isn't black and white so how can we expect to accurately encode that with 0's and 1's.

Like

©2022 by sophialau

bottom of page