Blog 5: Data & Society Databites Talk
Published on:
Video Source:
Databite No. 156: Hierarchy | Generative AI’s Labor Impacts
In today’s segment, I will be discussing the impacts on GenAI on labor based on the DataBite’s talk about the hierarchy in AI industry’s labor force.
The talk brought together subject matter experts in the AI industry, poeple who have a professional experience working with GenAI or who have interacted with the behind-the-scenes workers who are making AI what it is today. The first speaker is Mila Miceli, a researcher whose focus is on labor conditions and power asymmetries in the outsourced data workers examining their impact on ML and data sets. Her perspectives come from years of engagement with privatized data workers from the global south. The second speker is Russell Brandon, a tech editor for the Rest of The World news outlet. His background is in reporting tech news and writing special segments on emerging tech. Last but not least, we have John Lopez, a writer and filmmaker. His perspectives come from somebody whose work heavily collides with what GenAI is trying to accomplish. A lot of his and colleagues’ data has been used to train AI systems without accountability from the AI companies.
The overall aim of the talk is to give voice to those who are already doing the work in making GenAIa reality and those most impacted. The speakers really help us understand better what has led to the current of state of AI and its workforce and what invisible systems and hierarchies are scaffolding the situation and what might be coming next.
An example was brought up that highlights the power asymmentries in the AI industry. Data workers are only paid just enough to make ends meet and nothing more so they become highly dependent on the job. On the other hand, Engineers are making an absurd amount of money from the AI boom. This unfair wage distribution can be attributed to how society has come to glamorize AI engineers and forget about the data workers which point to ingrained stereotypes at play. In reality however, the AI engineers highly depend on the work of data workers in oreder to have something to work with.
Even though data workers have not always been at the forefront of conversations about AI, they have started gaining popularity in recent discourse similar to the one this blog is based on. The reason for this is there is a current shift towards looking for more professional data workers like writers, journalists, etc. because the AI systems are getting more and more specific and purpose oriented and they need to function in multiple languages. Th elevated functionality of AI systems makes talking about data a worthy discussion nowadays. Adding onto the current shift in AI functionality is also the need for AI companies to find workarounds of criticisms from artists and writers. If they hire the writers, they are now justified to use whatever work they produce as the companies’.
In the talk, one of the speakers goes on to critique one of the AI around, Copilot, on if it really helps with productivity like all AI promises to. Although Copilot helps to sumamrize meetings, hence minimizing the work middle management, does that really help companies be productive or does it just make them be able to be in more meetings since it is now easier, therefore continuing the unproductive cycle. This criticism really resonated with me as a casual user of AI for text generation and editing. More often than not, the texts I generate using AI do not reflect my personal voice and I have had to spend a significant amount of time, after the text has been generated, to edit the content myself which almost ends up making me think if I would have been better off writing the text myself from scratch.
As an emerging technologist, one of the points that touched me the most is the unfair wage distribution in the AI industry labor force. Even though we all aspire to live our best lives and have the best jobs to support those lives, that dream usually gets realized at the expense of others who have to take an insanely low check just because they have no other choice. I think it is the role of each and every technical engineer to advocate for those who are not so unfortunate as to negotiate a fair wage that reflects their contribution to society. Especially if those unfortunate ones enable us to do our job faster and better.
To end this blog, I had a question that I would have loved to ask the speakers had I gotten a chance to. DO you think the issue of unfair wage distribution warrants a need for a truly Universal Basic Income since AI is already uniting people from all over the world(data workers in the global south and AI engineers in the west)towards a common goal of advancing AI? What prompted me to start pondering on this question is how I think there needs a deep structural change in order to address the issue of unfair wage distribution.