The Hybrid Intelligence Manifesto: A Multi-Level Socio-Technical Framework for Keeping Human Judgment Economically Indispensable
Citation. Sherson, J., Rafner, J., Bang, J. B., et al. (2026). The Hybrid Intelligence Manifesto: A Multi-Level Socio-Technical Framework for Keeping Human Judgment Economically Indispensable [Preprint]. Hybrid Intelligence.
Summary. This public manifesto advances Hybrid Intelligence (HI) as a socio-technical design program intended to keep contextual, accountable human judgment economically central as AI becomes more agentic. Drawing on more than 50 interdisciplinary roundtables and two foresight workshops, it proposes definitions, classification instruments, use-case mappings, and scenario analysis rather than a comparative test of whether hybrid workflows outperform automation. The manuscript’s core claim is that complementarity is a dynamic, designable frontier between machine-accessible prediction and human judgment grounded in contextual knowledge that has not been operationalized for the system.
Key claims
- The manuscript defines human-AI complementarity as a dynamic, situational prediction-judgment frontier determined by which contextual knowledge has been externalized and made operationally available to machine systems in a specific socio-technical setting.
- Its operational tests assign subtasks toward machine prediction when relevant knowledge is codified, performance is stable under realistic domain shifts, uncertainty is calibrated, and errors are detectable and tolerable; they retain human judgment where work depends on tacit, relational, firm-specific, contested, evolving, or accountability-bearing context.
- The proposed six-level interface classification distinguishes autonomous AI, casual and interactive AI, and three HI configurations—accelerating, empowering, and collaborative HI—with the latter three requiring observable structural support for human judgment.
- The manifesto argues that regulation-first governance can preserve symbolic oversight without embedding human judgment in value creation, and that hybrid trajectories require aligned interface, organizational, market, and educational investments that may initially underperform automation-first metrics.
Evidence & limitations
- This is a public July 2026 draft/manifesto, not a peer-reviewed empirical study; its framework is grounded in interdisciplinary roundtables, foresight workshops, selected cases, and companion-work citations, so its normative and causal claims require independent comparative and longitudinal testing.
- Public provenance is strong: the official Hybrid Intelligence landing page hosts a PDF with the exact same SHA-256 hash, page count, and extracted text as the ingested Drive file.
- The source distinguishes demonstrated buildability and measurement of selected HI mechanisms from its untested wager that granular hybrid workflows can match or exceed automation-first architectures while reducing dependence on frontier-scale compute.