Humanistic Inquiry and Critical AI Research: Contexts, Convergences & Connections

The guest lecture titled “Humanistic Inquiry and Critical AI Research” delivered by Dr. D. Roy offered a comprehensive and interdisciplinary overview of the socio-cultural, political, and historical contexts surrounding contemporary artificial intelligence (AI). The lecture underscored the urgent necessity of humanistic inquiry in understanding the implications of AI systems, arguing for a critical perspective grounded in cultural studies, digital humanities, and media theory.

Dr. Roy began by mapping the evolution of artificial intelligence across different hype cycles, from the early enthusiasm in the 1950s to the ‘nuclear winter’ of the 1980s, and finally to the current resurgence driven by neural networks and deep learning models. The rise of deep learning post-2008 was attributed to two major developments: the explosion of data via the World Wide Web and the increased computational capacity provided by graphical processing units (GPUs), notably developed by companies like Nvidia. The landmark year of 2012 was highlighted, with the introduction of the ImageNet dataset by Fei-Fei Li and the breakthrough achieved by AlexNet, which achieved near-human accuracy in image recognition.

However, Dr. Roy critically questioned the mythologizing of AI as “intelligent” or “learning.” Drawing on thinkers such as Dan McQuillan and Meredith Whittaker, he emphasized that AI systems primarily perform statistical optimization tasks—what he termed “mathematical minimization”—rather than any form of cognitive intelligence. He cautioned against the hype surrounding generative AI and presented sobering findings from the 2025 MIT Generative AI Report, which revealed that 95% of enterprise investments in AI yielded no significant returns. This, he argued, reflects the commodification of “AI” as a marketing term rather than a meaningful technological benchmark.

Through his discussion of algorithms as “black boxes,” Dr. Roy foregrounded the opacity and abstraction of AI systems. The complexity of neural networks and vast datasets renders these systems difficult to audit or fully comprehend, even by their creators. He proposed that rather than being seen as value-neutral, AI must be situated within its cultural, economic, and political contexts.

This is where humanistic disciplines, particularly digital humanities, come into the picture. Dr. Roy elaborated on how digital humanities, born from the tradition of humanities computing, provide robust methodologies for analyzing AI systems. He discussed “distant reading” and “cultural analytics” as critical tools that can complement traditional close reading in literary studies. Using examples from Franco Moretti’s work at the Stanford Literary Lab, he illustrated how network analysis can provoke new insights without displacing human interpretative authority.

The lecture traced the long history of humanistic inquiry beyond the Eurocentric printed text to ancient Indian epigraphy and manuscripts, arguing that today’s digital platforms are simply the newest mediums for age-old scholarly practices. The concept of “text” was expanded to encompass networks, with algorithms themselves understood as forms of textuality. This reconceptualization enables a humanistic interrogation of algorithmic bias, drawing attention to embedded human and systemic biases in data collection and model training.

Dr. Roy also critiqued the exploitative labor underpinning AI infrastructures. From data annotators in Kenya earning meagre wages to the underreported reality of human surveillance in so-called “automated” retail systems, he illuminated the hidden human cost of AI. Referencing books such as Ghost Work and Data Feminism, he advocated for a justice-oriented framework that recognizes the ethical and socio-economic dimensions of AI development.

Towards the end of the lecture, Dr. Roy introduced his own collective, Critical Lens on AI in Majority Worlds, which aims to promote critical AI research grounded in local and global South perspectives. He distinguished between global North approaches that prioritize fairness and accountability, and global South approaches that emphasize data justice and post/decolonial computing. Both paradigms, he stressed, are essential and context-dependent.

Finally, Dr. Roy argued that literary and cultural scholars are uniquely positioned to contribute to critical AI studies. With their training in contextual analysis and interpretative methodologies, humanists can play a crucial role in shaping the ethical deployment and understanding of AI systems. He concluded with a call for interdisciplinary engagement, reminding the audience that AI is not merely a technical artifact but a socio-technical assemblage deeply entangled with structures of power and meaning.