Our research is source-attributed and open to correction. Our forecasts are questions under development, not certified predictions. Here is how to distinguish evidence from interpretation.
Named publisher: Cortist editorial project. No invented journalist credentials, endorsements or expert review.
Topic pages explain an AI subject and its context. Forecast questions specify a future outcome and a deadline. News reports a dated development. Methodology defines our process. These content types are linked but are not interchangeable.
Each substantial factual claim should point to a primary document when available. Supplier announcements and internal benchmarks are attributed to the reporting organization. Independent replication is labeled separately and never implied merely because a paper is published.
Cortist is an early-stage AI-assisted editorial project. Drafting and research assistance do not establish expert, regulatory or independent review. We disclose evidence boundaries and do not invent expert endorsements or a statistically proven forecasting record.
Research benchmarks, clinical studies, approvals and patient outcomes are different levels of evidence. Comparators, sample sizes, funding and limitations should be displayed when they materially affect the claim.
The current research overview covers 13 AI topic questions, including eight published numerical editorial drafts (F-001–F-008). Separately, two publisher-approved milestone forecasts (CM-003 and CM-004) were locked at their first public release on 27 October 2026. The eight drafts remain unscored; the two official records are also unresolved and unscored. None is externally reviewed or statistically calibrated. Provisional working percentages must not be called calibrated probabilities. The two original JSON files and their SHA-256 hashes can be checked through the public forecast integrity record; the publishing receipt is not an independently witnessed timestamp.
Readers can challenge a factual claim or ask for a source correction through the public contact page. Material changes should retain the original publication date, add a correction note with its date and what changed, and preserve archived underlying forecast versions. Corrections must not silently rewrite a previously locked prediction.
These public sections answer different editorial questions and should not repeat each other.
How source claims, evidence types and corrections are handled across Cortist.
Prospective resolution rules, deadlines, editorial locks and Brier scoring criteria.
The outstanding AGI and ASI review gates and missing independent signatures.
Exploratory A/M/E assumption scenarios: illustrative, not locked or calibrated.
The internal HOLD findings and blockers before outside statistical review.
Only independently resolved and scored outcomes; there are currently none.
