SuperintelligenceMONITOR
Published by Center for Artificial WisdomLatest newsletter: Issue #1 · draft v1.11

Draft forecast · unresolved

General research autonomy will remain unestablished in the open evidence

We expect public evidence to remain insufficient for the broad claim below, with low qualitative confidence. A negative finding may reflect limited evaluation access rather than a capability limit.

Question version 1.0 · Rubric SI-1.0 · preservation baseline October 15, 2026 · draft and unscored.

Information cutoff. October 8, 2026. Primary sources reviewed on this date; later evidence is excluded from the initial expectation.

Timing window. October 9–December 31, 2026

Why this is plausible. The available fresh end-to-end evaluation is narrow and difficult, while the game result leaves research transfer untested. The five-task rule is our declared operational test of this evidence claim, not an agreed definition of general research autonomy. Rapid model changes, evaluation access, and unpublished results make the expectation fragile.

What this question tests

This question tests a defined public-evidence threshold. Its five-task rule is an editorial choice, not an established definition of general research autonomy. A lack of qualifying evidence cannot establish that all private systems lack the capability.

Reading note added October 17, 2026. The original wording and criteria below are unchanged. A substantive revision would require a linked successor question.

Success and failure criteria

The skeptical forecast is overturned if an evaluator independent of the developer publishes reproducible success on at least five fresh, preselected ML research tasks, matching or exceeding documented human innovations under declared comparable resources, with no substantive human interventions after launch. Novelty, training-data exposure, failed runs, code, and outcome verification must be reported.

Resolution process

Met if no qualifying public evaluation is found by December 31 after a documented review of the named primary sources and relevant papers; not met if a qualifying evaluation appears. Inadequate provenance or disputed novelty is ambiguous. This resolves an evidence claim, not the unknowable state of all private capabilities. Human editorial adjudication is required.

Review trigger. December 31, 2026, or a qualifying research evaluation.

What would weaken the forecast. A qualifying independent cross-task result would overturn the skeptical assessment even if earlier forecasts were correct.

Original forecast and performance

The question wording, resolution criteria, and original expectation are preserved. The October 17 prose revision does not revise this forecast.

Initial draft · 2026-10-08. No revisions or resolved outcomes. We use qualitative confidence because numerical probabilities and likelihoods are not justified here. This question cannot establish probability calibration.

Outcome interpretation will be disclosed by the Center for Artificial Wisdom. An independent reviewer is preferred; no separate adjudicator has yet been assigned. These drafts are not included in performance claims.

Primary evidence informing this expectation

Read Issue #1