Does Morrison's distinction between defensive and acquiescent silence predict differential AI adoption behaviors? Are there empirical studies in high power-distance cultures (PDI > 60) on how silence norms interact with peer influence as an adoption mechanism?
How does verification workload affect the development of expertise and implicit cognition in human-AI interaction? Does heavy verification demand slow or prevent the development of AI-specific tacit knowledge?
Does the class/register advantage documented in chat-mode prompting persist or amplify when users must specify multi-step agent workflows rather than compose individual prompts?
Is there evidence that crisis-driven centralization degrades the distributed information environment in ways that make AI deployment more likely to fail silently? Who catches AI errors in organizations where upward honesty has been structurally suppressed?
Does a failed or over-promised AI initiative reduce the probability of success for subsequent initiatives in the same organization? Any work that models organizational trust as a depleting resource rather than a stable background condition?
In contexts where AI distributes causality, do existing feedback inequalities (junior staff bearing disproportionate blame) compound or attenuate?
How do agentic or automated AI systems foreclose user-level tactical adaptation? Does platform dependency and deskilling literature apply to LLMs or generative AI?
What does it cost (in knowledge codification, documentation, explicit constraint-setting) to make a knowledge-intensive professional workflow legible to an AI agent — and who bears that cost?
Does AI tool use in already-surveilled workplaces accelerate diffusion of responsibility, or does it provide accountability artifacts that restore it?
Is there evidence for a necessary 'slow phase' of direct model interaction before effective use of more autonomous agentic systems? What happens when users skip foundational interaction phases?
How crisis-driven centralization systematically destroys the distributed information environment it was designed to fix, and why emergency authority rarely reverts.
Why more than a third of feedback interventions degrade performance, and how power-distance culture and defensive silence compound that failure.
Why adoption rate figures vary 5–40 points depending on method, and why peer-to-peer influence outperforms formal training as an adoption mechanism.
Why the constraint in AI-augmented work shifts from processing capacity to human judgment bandwidth, and what memory architecture agents need to stop failing silently.
How working with AI shifts thinking from vertical hierarchical knowledge to rhizomatic horizontal fields, and why opaque reasoning chains make knowledge transfer impossible.
How LLM performance gaps map onto Bourdieu's linguistic capital, making AI the most efficient class-sorting machine ever built.
Whether consciousness is metabolically expensive, evolutionarily unnecessary, and possibly a costly side-effect rather than the engine of intelligence.
How Diminishers create cognitive ceilings that cap organizational performance, and the mechanisms by which Multipliers extract twice the available intelligence.
How the four-year degree became a cheap filtering proxy rather than an ability signal, the compounding cost of that convenience, and IBM's pivot toward aptitude-based recruitment.
Why individual empathy is not the problem — the institution is — and the underdeveloped prescription for building organizations that don't select against it.