GOVERNATLAS / PRODUCT DOSSIER

AI Explainability 360

The AI Explainability 360 project documents an open-source toolkit for explaining datasets and machine-learning models.

Explore the product ↗Add to comparison ↗

01 / CONTEXT & SCOPE

Understand the proposition.
Define the boundary.

Where it fits

Use its method taxonomy to select an explanation appropriate to the reader, data and review question.

The decision to make

Use the explanation taxonomy to choose a method rather than treating a collection of algorithms as one assurance claim. Local, global, data and model explanations have different consumers and assumptions. The README also points to a separate newer in-context toolkit; do not transfer that toolkit’s LLM functionality to this profile.

What this record does not establish

The toolkit is distinct from AI Fairness 360 and the separately referenced ICX360 project. Proxy explanation metrics, available algorithms and sample notebooks do not establish causal validity, fairness or current compatibility of every optional dependency.

1 current source · 4 published assertions
Source coverage is a research boundary, not a quality score.

02 / YOUR EVALUATION

Turn interest into a useful trial.

Open the evaluation playbook ↗

A test worth running

Choose a synthetic model and one explanation question, pin the selected algorithm and its dependency environment, and retain the explanation output. Ask a second reviewer to interpret it using only the manifest, then record an ambiguity that requires follow-up rather than declaring the model explained.

GovernAtlas editorial proposal. We have not performed this test.

Evidence to bring back

A chosen algorithm/reader/question/dependency manifest.
An explanation artifact with reviewer interpretation and unresolved limits.

THE OPERATING DETAIL

Choose the explanation before choosing its presentation.

GovernAtlas analysis: practical review considerations, not findings from a product test.

Identify the intended reader

Write down whether a developer, operator or model reviewer needs the explanation and what they must do with it. Distinguish a data summary, local prediction explanation and global model explanation. A method suited to one task should not be repurposed solely because its visualization looks persuasive.

Examine the proxy metric

For one synthetic explanation, retain the selected proxy metric, input, parameters and result. Ask what property the metric measures and what it omits. Compare that answer with the actual reader’s question. A favorable faithfulness or monotonicity result does not establish every aspect of explanation usefulness.

Keep toolkit boundaries explicit

Pin the AIX360 release and supported environment for the selected algorithm. Treat the separate ICX360 reference as a different toolkit requiring its own review. Shared research ownership and similar names do not establish that an LLM explanation capability is available in the AIX360 implementation being installed.

03 / DOCUMENTED EVIDENCE

Read the claim. Inspect the basis.

Each statement retains its classification and preserved source. A vendor assertion is not an independent test result.

first party asserted

Its listed proxy explanation metrics include faithfulness and monotonicity. The README points to a separate ICX360 toolkit for LLM input explanation; that scope is not attributed to AIX360.

Source, dates & preserved provenance

AI Explainability 360 — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 3f1fc27b95f3b63cab0592cf18cbca6b1229da96aa566f6207b1d0d9d174e5df

The hash identifies the preserved capture. The live source may have changed.

first party asserted

Its taxonomy distinguishes data and model explanations, direct and post-hoc methods, and local and global views.

Source, dates & preserved provenance

AI Explainability 360 — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 3f1fc27b95f3b63cab0592cf18cbca6b1229da96aa566f6207b1d0d9d174e5df

The hash identifies the preserved capture. The live source may have changed.

first party asserted

AIX360 documents explainability algorithms and proxy metrics for tabular, text, image and time-series data.

Source, dates & preserved provenance

AI Explainability 360 — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 3f1fc27b95f3b63cab0592cf18cbca6b1229da96aa566f6207b1d0d9d174e5df

The hash identifies the preserved capture. The live source may have changed.

first party asserted

The AI Explainability 360 project documents an open-source toolkit for explaining datasets and machine-learning models. Scope: primary documentation only; no independent model test, fairness determination, causal finding or compliance approval.

Source, dates & preserved provenance

AI Explainability 360 — primary scope and current identity ↗

Authority: first party
Observed 2026-10-16 · Reviewed 2026-10-16 · current until 2026-11-15

Artifact SHA-256: 3f1fc27b95f3b63cab0592cf18cbca6b1229da96aa566f6207b1d0d9d174e5df

The hash identifies the preserved capture. The live source may have changed.

Source register

AI Explainability 360 — primary scope and current identity ↗first party · captured 2026-10-16

Brand asset provenance

Organization identity asset from its official website; inclusion does not imply endorsement.

Official website ↗ · Original asset ↗

04 / BEFORE YOU COMMIT

Better questions.
A clearer decision.

Resolve these with the provider and your own evaluation team.

Can you provide a chosen algorithm/reader/question/dependency manifest for the selected configuration?

Can you provide an explanation artifact with reviewer interpretation and unresolved limits for the selected configuration?

Which exact version, entitlements, data paths and responsible roles apply?

Regulatory applicability requires a separate, current determination. Inclusion is not certification.

COMPLIANCE / CLAIMS, SCOPE & GAPS

Start with the requirement.
Inspect the support.

Prepare your vendor review ↗

These labels describe this dossier’s current evidence coverage. A published mention may be a framework mapping, a qualified assertion or a limitation; it is not a compliance finding.