When thinking about human-centered AI and automation, it is imperative to be mindful of the labor that goes behind the commercial proliferation of large-scale AI systems. As technologies like large language models, text-to-image models and other forms of generative AI are rapidly introduced within a number of production pipelines, in a post-pandemic reality, we need more academic interest on the potential devaluation or displacement of human labor as a consequence, and what can be done to mitigate that.
Recent literature has attempted to recognize human interventions at different stages of the commercial AI lifecycle, starting from data preparation and software development to the often hidden labor associated with ‘patchwork’- calibration and troubleshooting of AI technologies deployed in the market. Prescient inquiries have also been made on the potential encroachment of models like MidJourney, DALLE or Stable Diffusion upon creative labor. In this proposed session, we attempt to raise further questions on whether tolerability around plagiarism or copyright for AI produced art or products need to be redefined. We also seek to inquire about the requirement of a robust economic model to properly remunerate workers whose labor gets crowdsourced to train and fine-tune AI systems that seek to replace them.
Since the current generation of educators influence the next generation of researchers, it is also essential to initiate dialogue with them around how principles of ethical labor can be integrated into AI pedagogy and learning and introduced early on in the academic journey of AI and data practitioners.