Dataiku and Harris Poll, 800 data leaders surveyed across 7 jurisdictions:
95% cannot trace AI decisions end-to-end.
5% could supply that trace to a regulator 100% of the time.
80% say accurate-but-unexplainable AI is more dangerous than wrong-but-traceable.
The substrate that closes the gap is filed at the UK IPO. 31 applications, 914 claims, four weeks before the report was published.
https://mickai.co.uk/articles/the-95-percent-gap-dataiku-confessions-mickai-substrate
Eight hundred data leaders, surveyed by Dataiku and The Harris Poll between August 2025 and February 2026, admit a structural failure that British engineering has already answered. Ninety five per cent cannot trace AI decisions end-to-end. Eighty per cent say accurate but unexplainable AI is more dangerous than wrong but traceable. Seventy two per cent allow agents to make critical decisions without explanation. Mickai's thirty one filed UK patent applications, nine hundred and fourteen claims, named inventor Micky Irons, are the engineering primitives that close every gap the report enumerates.
ITmedia AI＋ (@itm_aiplus)
전 OpenAI 임원이 비영리 AI 감사단체 'AVERI'를 설립했다. 목표는 소비자 가전 수준의 표준화된 안전 점검(가전정도의 안전확인)을 통해 AI 시스템의 안전성을 검증하는 체계를 구축하는 것이다.
💬 P.S: For readers working in the Dutch context: we are looking for organizations who would like to participate in a one-year pilot (starting early 2026) to test our AI performance appraisal instrument in practice.
📩 Interested? Contact us at [email protected].
💻 La IA redefine la web: @Hostinger_ES presentó #Web2Agent, #AIAudit y #Kodee para adaptar sitios y servicios a la nueva era digital. Más del 70% de los usuarios mexicanos ya usan herramientas impulsadas por IA. #NoticiasMX
Tony Cantero Suárez
Mickai
sayzard
Data School
Sociedad Noticias