AI isn’t Artificial or Intelligent: For the past seven years, Krystal Kauffman has worked for Turk. She works for Mechanical Turk (MTurk), an Amazon-owned crowdsourcing platform that enables companies to employ employees to carry out different activities in exchange for payment. According to Kauffman, many of these challenges have been used to train AI programs.
“We’ve worked on a number of significant AI projects in the past. In order to train AI to recognize different voices and things, there are tasks where users only need to repeat a phrase six times. Kauffman explained this to Motherboard. Therefore, I get to handle a little bit of everything, although there is now a lot of machine learning and AI data labeling. There has been a rise in the number of requesters who list the work.
‘Use Any App But Stay Away From WhatsApp’ – Telegram Founder Pavel Durov Warns Users
In addition to searching through vast datasets to identify photographs, filter NSFW content, and describe objects in images and videos, Kauffman is a member of the massive work force that powers AI. For data training firms like iMerit, Sama, and Alegion, these repetitive and unappealing jobs are frequently outsourced to gig workers and employees who primarily reside in South Asia and Africa. One of the most cutting-edge algorithmic content moderation systems on the internet, for instance, is found on Facebook. However, the ‘artificial intelligence’ that powers that system is ‘learning from thousands of human judgements’ made by human moderators.
The creation of AI is at the heart of the activity of major corporations like Meta and Amazon, which both have strong AI development teams. Our capacity to effectively use the most recent AI technology at scale will determine where we are in the future, according to Meta, and Amazon urges users to “innovate quicker with the most complete collection of AI and [Machine Learning] services.”
“We’re used to working on things that we just don’t know exactly what they are for […] we know we’re working on some of those big tech devices. And while I don’t need to be called an employee or anything like that, you very rarely hear a big tech company acknowledge the invisible workforce that is behind a lot of this technology.”
The largest IT corporations in the world envision a time soon when AI will replace most human labor and unleash new levels of productivity and efficiency. However, this vision misses the fact that a significant portion of what we currently refer to as “AI” is actually powered by laborious, underpaid human work.
“One of the misconceptions of AI computing, in my opinion, is that it actually operates as planned. According to Laura Forlano, associate professor of design at the Institute of Design at Illinois Institute of Technology, “I believe that human labor is now filling in a lot of gaps in the way that the systems work. “On the one side, the business can say that these things are, you know, miraculously happening behind the scenes or that most of what’s happening is the computer. However, we are aware that in numerous instances, including the creation of online content, the control of autonomous vehicles, and the manufacture of medical equipment, human labor is being used to fill in the gaps where automation falls short.
“We’re used to working on items whose exact purposes we are unsure of; we are aware that we are developing some of those high-tech devices. And even if I don’t need to be named an employee or anything, Kauffman noted that it’s uncommon to hear a major tech corporation acknowledge the unseen workforce that is responsible for much of this technology. They mislead the public into thinking that AI is more sophisticated and intelligent than it actually is, which is why we must continue to teach it daily.
Tens of thousands of gig workers are employed by tech companies to keep up the appearance that their machine-learning algorithms are entirely autonomous and that each new AI product is capable of automatically resolving a variety of problems. Contrary to popular belief, the process of developing AI shares a great deal more in common with that of manufacturing.
“I don’t believe that the general people is very conscious that there is a supply chain. The global supply chain has asymmetrical geographic flows and relationships. A postdoctoral researcher at the Oxford Internet Institute working on the Fairwork project, Kelle Howson, told Motherboard that the project is based on a significant amount of human labor.
Howson can’t tell for sure whether tech corporations are concealing human AI workers on purpose, but it is clear that doing so serves their interests. Since there is an impression that the work is over, Howson believes that doing so promotes their business models in various ways. “As a client, you log into the platform interface, post your project, and the work is promptly delivered. It almost seems magical. There is a perception of efficiency since it appears as though no humans were ever involved. And that actually fits with the storylines Silicon Valley prefers to tell. the ideas of disruption, technological solutionism, and swift destruction of established order.
The AI pipeline is severely unbalanced, just like other international supply chains. By performing frequently low-paying beta testing, data annotation and labeling, and content moderation tasks, developing nations in the Global South power the development of AI systems, while centers of power in the Global North profit from this work.
