AI Sprints: Towards a Critical Method for Human-AI Collaboration

2025-12-13

AI Summary

This study addresses how large language models (LLMs) are transforming research paradigms in the humanities and social sciences by proposing “AI Sprints”—time-bound, intensive, human-AI collaborative research methods. Methodologically, it introduces an original operational framework integrating critical reflection with iterative generative AI dialogue, synthesizing data sprints, book sprints, and critical code studies, while extending Rogers’ digital methodology to develop an AI-native protocol analysis pathway. Theoretically, it identifies three cognitive modes—“cognitive delegation,” “productive augmentation,” and “cognitive load”—and constructs an ethics-computation dual-track framework. Practically, it yields a reusable AI-augmented research workflow and establishes collaborative principles safeguarding scholarly agency, thereby offering digital humanities a novel paradigm that advances both methodological innovation and epistemological depth.

Technology Category

• Humans and AI: Understanding People, Theories, Concepts and Methods
• Application Domains: Humanities & Computational Social Science
• Philosophy and Ethics of AI: Artificial General Intelligence

Application Category

• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

The emergence of Large Language Models presents a remarkable opportunity for humanities and social science research. I argue these technologies instantiate what I have called the algorithmic condition, whereby computational systems increasingly mediate not just our analytical tools but how we understand nature and society more generally. This article introduces the possibility for new forms of humanistic inquiry through what I term 'AI sprints', as intensive time-boxed research sessions. This is a research method combining the critical reflexivity essential to humanistic inquiry with iterative dialogue with generative AI. Drawing on experimental work in critical code studies, I demonstrate how tight loops of iterative development can adapt data and book sprint methodologies whilst acknowledging the profound transformations generative AI introduces. Through examining the process of human-AI collaboration when undertaken in these intensive research sessions, I seek to outline this approach as a broader research method. The article builds on Rogers' digital methods approach, proposing that we extend methodologies to study digital objects through their native protocols, using AI systems not merely to process digital traces but to analyse materials traditionally requiring manual coding or transcription. I aim to show this by introducing three cognitive modes, cognitive delegation, productive augmentation, and cognitive overhead, explaining how researchers can maintain a strategic overview whilst using LLM capabilities. The paper contributes both a practical methodology for intensive AI-augmented research and a theoretical framework for understanding the epistemological transformations of this hybrid method. A critical methodology must therefore operate in both technical and theoretical registers, sustaining a rigorous ethical-computational engagement with AI systems and outputs.

Problem

• Develops AI sprints for humanistic inquiry with AI
• Proposes a method combining critical reflexivity and AI dialogue
• Explores cognitive modes for strategic AI-augmented research

Innovation

• AI sprints combine humanistic reflexivity with generative AI dialogue
• Method adapts data and book sprints via iterative human-AI loops
• Uses AI to analyze materials needing manual coding or transcription