Care, Classification, and the Politics of Who Answers
Johan Michalove2026-02STSInfrastructure Studies
This paper examines the emergence of voice-based artificial intelligence in social services through three analytical lenses drawn from science and technology studies: infrastructure, care, and testing. Drawing on Susan Leigh Star and Karen Ruhleder's relational definition of infrastructure, I ask not what a voice AI system is but when it becomes an infrastructure of care—and for whom.
Evidence-Based Conversational Architecture for Voice AI in Community Resource Navigation
Johan Michalove2026-02HCIDesign
This paper presents an evidence-based design methodology for voice AI in social service information and referral, synthesizing principles from six domains: the history of telephone operators, crisis hotline best practices, IVR usability research, 211/311 systems, telephone-based social services, and social work intake models. We derive 26 design principles treating the voice interaction as a care relationship rather than an information retrieval task.
Johan Michalove2026-02Critical ComputingSTSAI Ethics
A critical reflection on the experience of building artificial intelligence for community care during a period of political crisis, technological acceleration, and state violence. It is not a technical paper; it discloses no architecture, no methods, no implementation details. It is a paper about stance: the ethical, political, and intellectual commitments that precede and condition the work of building.
The minimum viable knowledge to begin contributing to cancer bioinformatics: what cancer is, how it is classified, how it is treated, where computation enters, what the key datasets are, what the open problems are, and where to start. Opinionated. Assumes you learn by building.
A Literature Review for Bioinformaticians, Systems Engineers, and the Impatient
Johan Michalove2026-02Cancer BioinformaticsAI
A survey of computational cancer research across eight domains: genomic variant interpretation, single-cell and spatial transcriptomics, digital pathology, protein structure and drug discovery, liquid biopsy and early detection, large language models in clinical oncology, laboratory automation, and data equity. The field's most consequential challenges are no longer primarily algorithmic but infrastructural.
Hans Christian Andersen, arranged by Johan Michalove2026-02Design
The final scene of Andersen's 1845 story, rendered as a triptych: English and Danish in parallel columns, Hans Tegner's plates in the margins. A threshold text. The girl does not cross into the new year alive. What follows from that is the question this paper will answer.