AI2030 Responsible AI Tools Marketplace

Key Takeaways

XAI methods need a unified framework to be trustworthy: The paper argues that different explainability methods (like attribution, counterfactual, and concept-based approaches) draw on different types of information and make different assumptions, but there's been no common way to understand how they work. GFGE provides this unified framework, which is essential for ensuring explanations are reliable and comparable across different AI systems.
Explanations are a two-way communication process, not just technical outputs: The framework emphasizes that explanations involve both "sense-reading" (the AI analyst interpreting system evidence) and "sense-giving" (communicating that interpretation to users). This means a good explanation depends on matching the right information to the right audience—a technical explanation for engineers differs from one for regulators or affected individuals.
Tracing assumptions and limitations is critical for responsible AI: GFGE operationalizes the practice of keeping "evidence records" that retain the sources, assumptions, and limitations of explanatory claims. This transparency about what we don't know and what we're assuming is fundamental to responsible AI deployment, particularly in high-stakes domains like healthcare or criminal justice.
Different XAI methods can coexist within one framework: Rather than declaring one explanation method superior, GFGE shows how intrinsic methods (built into the model), post-hoc methods (applied after), and hybrid approaches can all be represented through the same five roles while preserving their distinct evidential requirements. This supports more nuanced, context-appropriate choices about which explanations to use.
Context and audience should drive which explanations you use: The framework emphasizes that "the explanatory question, audience, and context guide the procedure," meaning there's no one-size-fits-all explanation. Responsible AI requires choosing explanation methods deliberately based on who needs to understand the

Abstract

Explainable artificial intelligence (XAI) encompasses methods that draw on different sources of information and address different explanatory needs. A common framework is needed to describe how this information becomes evidence and is communicated as an explanation for a particular recipient. We propose the General Framework for Generating Explanations (GFGE), grounded in interpretative frameworks and the complementary activities of \emph{sense-reading} and \emph{sense-giving}. Its conceptual foundation is the Interpret/Explain Schema (IES), which connects an analyst's interpretation of system evidence, the communication of a selected account, and the recipient's interpretation of that account. GFGE operationalises this schema through five roles: data interpretation, model interpretation, output interpretation, optional post-hoc analysis, and aggregation. A role-typed operation graph records method-specific dependencies, while evidence records retain the sources, assumptions, and limitations of explanatory claims. The explanatory question, audience, and context guide the procedure. We instantiate GFGE for attribution, surrogate, counterfactual, concept and prototype, intrinsic rule, argumentation, and language-model methods. These instantiations show how intrinsic, post-hoc, and hybrid workflows can be represented through the same roles while preserving their distinct evidential requirements. GFGE provides a common basis for analysing explanation workflows, tracing communicated claims to their evidence, and identifying unresolved explanatory dependencies.

Key Takeaways

XAI methods need a unified framework to be trustworthy: The paper argues that different explainability methods (like attribution, counterfactual, and concept-based approaches) draw on different types of information and make different assumptions, but there's been no common way to understand how they work. GFGE provides this unified framework, which is essential for ensuring explanations are reliable and comparable across different AI systems.
Explanations are a two-way communication process, not just technical outputs: The framework emphasizes that explanations involve both "sense-reading" (the AI analyst interpreting system evidence) and "sense-giving" (communicating that interpretation to users). This means a good explanation depends on matching the right information to the right audience—a technical explanation for engineers differs from one for regulators or affected individuals.
Tracing assumptions and limitations is critical for responsible AI: GFGE operationalizes the practice of keeping "evidence records" that retain the sources, assumptions, and limitations of explanatory claims. This transparency about what we don't know and what we're assuming is fundamental to responsible AI deployment, particularly in high-stakes domains like healthcare or criminal justice.
Different XAI methods can coexist within one framework: Rather than declaring one explanation method superior, GFGE shows how intrinsic methods (built into the model), post-hoc methods (applied after), and hybrid approaches can all be represented through the same five roles while preserving their distinct evidential requirements. This supports more nuanced, context-appropriate choices about which explanations to use.
Context and audience should drive which explanations you use: The framework emphasizes that "the explanatory question, audience, and context guide the procedure," meaning there's no one-size-fits-all explanation. Responsible AI requires choosing explanation methods deliberately based on who needs to understand the

Abstract

GFGE: Unifying Explainable AI Methods through an Interpretation Framework

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links

GFGE: Unifying Explainable AI Methods through an Interpretation Framework

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links