AI for Science Strategic Compass (AFSC) – Full Strategy Matrix

AFSC Strategy Matrix. The AI for Science Strategic Compass (AFSC) is a function-based decision framework that helps scientists make strategy-level choices about how to use AI to address specific scientific discovery problems. It aligns six core AI functions (Represent; Reason & Infer; Optimise & Control; Simulate & Emulate; Generate & Create; Autonomise & Orchestrate) with four cross-domain, recurrent discovery tensions (System Complexity, Experimental Constraint, Data Scarcity, Combinatorial Explosion), yielding a 6×4 Strategy Matrix. In this matrix, rows are core AI functions and columns are discovery tensions; each function-tension cell shows how the function can mitigate the tension. For each function, the left-hand column shows its two intrinsic binary axes and the three resulting atomic categories: a minimal triad that is mutually exclusive and collectively exhaustive. This triad provides a small, stable set of building blocks, so higher-level strategies can be expressed as combinations of these atoms. Each cell contains: (i) a concise keyword naming the function's mitigation logic for that tension; (ii) three strategic pathways capturing mitigation mechanisms, each anchored to a minimal atomic signature (a single atom or the smallest sufficient set); and (iii) representative method families for each pathway, indicating that the pathway is actionable at the method level; these families refer to existing strands of work in the literature and are illustrative rather than exhaustive. Stars (★) mark high-leverage cells—typical but non-exclusive entry points for that tension. AFSC turns a fragmented, fast-evolving AI landscape into a stable, problem-centred map for clear, defensible AI strategy choices.

Using the Matrix.
1. Identify the primary discovery tension(s) for your scientific problem.  
2. In the corresponding tension column, use the starred (★) cells as high-leverage entry points.  
3. From the chosen cells, select the strategic pathways whose mechanisms fit your data situation and required type of evidence.  
4. For each selected pathway, choose method families that realise that pathway in practice and match your local constraints.  
5. Instantiate these families with specific models, algorithms or workflows, and record the reasoning chain (tension → cell → pathway → family → method) so that your AI strategy is explicit, defensible, and revisable.

The six core AI functions and their atomic triads used in this matrix are defined in more detail in the AI Core Function Ontology: A Two-Level Capability Framework (DOI: 10.5281/zenodo.17664037), which provides the capability backbone for AFSC.

The full conceptual framing, validation of AFSC, formal definitions of the strategic pathways, and canonical references for method families are provided in companion research manuscripts. This record is limited to the high-resolution figure and a concise usage guide.