Course:ELEC300E/CaseStudies/AITrainingImpacts

Context: Data Colonialism

Chatbot Training in Nairobi
Image:wiki.png
Content / Topics
The Digital Divide
Digital Literacy
Participatory Design
Relevant Analysis Strategies or Tools
Decision Impact Assessment
Iceberg Model
Systems Mapping

A concept that you might find useful as you read this case study is the idea of "data colonialism".

Colonial systems "seek to impose the will of one people on another and to use the resources of the imposed people for the benefit of the imposer" (Asante, 2006). Throughout history, these resources have included labor, land and resources, but in recent years, and particularly with the advent of AI, data has become a valuable commodity.

Researchers introduced the term 'data colonialism' to describe how the commodification of data can lead to forms of exploitation that echo historic colonialism:

"We are often told that data are the new oil. But unlike oil, data are not a substance found in nature. It must be appropriated [...] Data relations enact a new form of data colonialism, normalizing the exploitation of human beings through data, just as historic colonialism appropriated territory and resources and ruled subjects for profit."

Case: Chatbot Training in Nairobi

Training Generative AI requires vast amounts of representative data. In some cases, this data already exists and can be bought or scraped from existing sources, but in other cases companies employ workers to help create training data. Many of these workers live in Africa, in particular in cities like Nairobi, where there are fluent English speakers. AI workers often work long hours and are paid low wages (1-2$/hour).

A recent CBC podcast tells the story of Micheal Geoffrey Asia, a working in Nairobi's AI workforce,

Micheal worked as a "chat moderator" for a company called New Media Services, where his job was, in short, to chat with people online. These transcripts were then sold to AI companies for use training chatbots. Micheal was not told which AI companies were buying this data, or what kinds of chatbots were being trained. He was also not told exactly who he was talking to, but his guess was that they were lonely people who had paid to for intimate conversations.

What was Micheal told? Using multiple accounts and identities, Micheal was instructed to adopt false personas and to keep users engaged:

"It was the objective of the company to ensure that you keep these people locked to the system, you make sure they subscribe because every conversation meant revenue for the big tech companies. So you are like supposed to ensure that you keep the conversation flowing, keep them locked to the internet, keep them attached to the screen, and ensure that they generate interest day in, day out." (Artificial Intimacy, Behind the Bot)

Micheal began to find that this unusual work started impacting his mental health. He was having extremely personal conversations all day without understanding who he was talking to, what the conversations were for, while having to impersonate different online personas. Later on, seeing a trained chatbot in action, Micheal recognized Kenyan sentence structures, vocabulary and speech patterns (for example, use of the word "delve"). Micheal explains:

"I love calling [AI] Africa intelligence because Africa has been developing these softwares for them. Africa has collected literally every data for them. Africa has been doing all the dirty work for them. So let's avoid the whole issue of imagining that this is magic. It's not magic. It will never be magic. This is human labour behind the screens." (Artificial Intimacy, Behind the Bot)

Recommended Readings and Resources

• ~13 minute podcast from CBC (5:30 - 18:15): https://www.cbc.ca/listen/cbc-podcasts/1353-the-naked-emperor/episode/16219467-e4-behind-the-bot.
  • Or read the transcript here: https://www.cbc.ca/radio/podcastnews/artificial-intimacy-episode-4-transcript-9.7249905, (search for Micheal Geoffrey Asia to find the section discussing his experience)

Discussion Questions

Here are a few questions to help you begin analyzing the issues discussed in this case study. The purpose of these questions is to prompt reflection and further consideration: there are no right or wrong answers!

• Personal response: Take stock of your immediate response to this case study: outrage, discomfort, boredom, confusion, frustration, hope, interest? How might your unique personal experiences and perspectives inform this response? How can you leverage this response in a positive way?
• Social and Historical Context: Would you consider the way AI companies employ workers like Micheal to be exploitative, or an example of data colonialism? Why or why not? Does this story echo any other case studies, or recent news articles you have read?
• Contradictory viewpoints and outcomes: Micheal's story touches on some of the negative aspects of the AI-based economy in Africa. Can you think of any positive interpretations or potential outcomes?
• Positive action: Can you think of any technological, structural or social changes that would address aspects of this case study that you might find concerning? What concrete actions would this change require?
• Interruptions and invitations: In an interesting article on postcolonial computing, researchers argue for: "sensitivity to how uneven power relations are enacted in design practice". What practices might designers employ to take into account (or interrupt) uneven power relations through design practice?
• Physical connections: Like Africa, Canada has a colonial history. What might 'data colonialism' look like in Vancouver?
• Positive Practices: What is one concrete action that you could take, or a practice that you could adopt in your personal, academic or professional life to address some of the issues discussed in this case study?