AI Wiki

Explore AI Wiki

Corrections

When we get something substantively wrong, we fix it and log the fix here; small copyedits live in each article's revision history instead. Found an error? Use the "Report issue" button on the article (no account needed). So far we have received 2 reports and reviewed 1, taking a median of 4.8 days. 1 is waiting, the oldest since September 26, 2026. That is our record rather than a promise. See the verifiability policy for how checking works.

321 articles were corrected in the last 30 days. Every correction note and resolved report is listed below, newest first. A fix applied to several articles with the same note is shown once, listing each article; "See the change" opens that version in the article's revision history.

  1. editorial

    Consolidate the duplicate Groq LPU article into the independently verified Groq Hardware article.

  2. editorial

    Remove the reciprocal Groq LPU link before consolidating that duplicate article into Groq Hardware

  3. editorial

    Fixed Logic Theorist attribution, SCS and HCII chronology, SCS department structure, Robotics Institute funding, Navlab autonomy scope, dated institutional figures, and invalid internal links

  4. editorial

    Fix Custom GPT access, privacy, and adoption claims

  5. editorial

    Fixed THAID date and NP-completeness scope; qualified missing-data, categorical, interpretability, probability, importance, complexity, and ensemble claims

  6. editorial

    Rebuilt AI Alignment around 25 authoritative sources; removed unsupported risk claims; fixed EU source fit and reference rendering.

  7. editorial

    Reframed Social Media as a broad socio-technical system; replaced unsupported claims with 34 explicit sources; corrected a 2026 Nature citation and reference rendering.

  8. editorial

    Replaced false overfitting, out-of-bag, importance, default, and benchmark claims; added 26 explicit academic and first-party sources

  9. editorial

    Fixed Walker specifications, financing classifications, FY2025 results, shipment ranking, dated company metrics, and promotional claims; added primary-source URLs and July 2026 deployment updates

  10. editorial

    Repaired 28 bibliography emphasis spans that rendered as literal Markdown markers; factual prose, citations, source text, URLs, internal links, and categories are unchanged.

  11. editorial

    Added Muse Image privacy and chatbot safety-governance coverage

  12. editorial

    Repaired visible reference-title emphasis while leaving prose, claims, citations, links, source text, and categories unchanged.

  13. editorial

    Repair Mask R-CNN source link and reference emphasis rendering.

  14. editorial

    Repaired 34 bibliography emphasis spans that rendered as literal Markdown markers

  15. editorial

    Added state-space and diffusion language-model architecture and objective coverage

  16. editorial

    Clarified a security-study term and tightened supporting citations for Transformers, BLOOM, and model supply-chain research.

  17. editorial

    Reconstructed OpenAI o1 from primary and independent evidence; corrected full o1 context to 200K; removed unverified architecture, parameter, codename, genealogy, reward, market, and successor claims; separated o1-preview, research o1, API snapshot, and o1-pro results; added exact evaluation protocols, safety-test scope, API lifecycle dates, independent benchmark caveats, and explicit sources.

  18. editorial

    Rewrites the article around the original 164-task Python benchmark, corrects pass@k provenance and score interpretation, separates later variants, removes unverified leaderboard claims, and replaces broken footnotes with verified numeric citations.

  19. editorial

    Distinguish AgiBot production milestones, shipment estimates, disclosed deployments, bounded research results, artifact-specific licensing, and product security support.

  20. editorial

    Reconstructed AWS as Amazon's cloud segment and service platform; corrected the 2025 operating-income share and Amazon/AWS capex boundaries; date-stamped 2026 segment and Region data; bounded the Anthropic investment to primary evidence; removed unsupported market, customer, capacity, project, catalogue, and forecast claims; and added explicit primary, academic, standards, government, regulator, security, architecture, and incident sources.

  21. editorial

    Distinguished open-source AI from open-weight and source-available releases; verified OSAID 1.0 components, licensing boundaries, reproducibility, safety, and EU AI Act exceptions; removed unsupported model rankings, download shares, forecasts, cost claims, and vendor comparisons.

  22. editorial

    Reconstruct LangChain with cutoff-pinned package versions, exact product boundaries, and sourced coverage of agents, tools, retrieval, memory, licensing, deployment, and security.

  23. editorial

    Rebuilt Agentic AI around documented definitional disagreement, classical agent foundations, verified architectures and benchmarks; removed unsupported adoption, market, ROI, current-leaderboard, and NIST-specific claims; added bounded security and governance evidence with canonical links.

  24. editorial

    Reconstructed LoRA's definition, mathematics, initialization, scaling, targeting, efficiency boundaries, original evidence, variants, QLoRA, serving, composition, reproducibility, and limitations; corrected the Prefix-Tuning comparison and bounded the NeurIPS 2025 spectral comparison; removed unsupported deployment, popularity, and universal-performance claims.

  25. editorial

    Rebuilds Learning Rate around assumption-bounded optimization theory, primary-source evidence, reproducible tuning guidance, correctly rendered schedules, and explicit limits on universal prescriptions.