推理聲明代幣(RCTs):外部表徵治理中可檢驗AI推理的觀察性建構

Hacker News·

本文介紹了推理聲明代幣(RCTs),這是AIVO標準內的一種新構建,旨在增強外部表徵中AI推理的可檢驗性。RCTs捕捉AI系統在推理過程中產生的離散、時間索引的推理聲明,從而促進面向治理的可追溯性,而不斷言其正確性。

Image

Info: Zenodo’s user support line is staffed on regular business days between Dec 22 and Jan 4. Response times may be slightly longer than normal.

Image

Reasoning Claim Tokens (RCTs): An Observational Construct for Inspectable AI Reasoning in External Representation Governance

Creators

Description

As large language models increasingly mediate external representations of enterprises, products, and services, governance risk has shifted from output quality toward post-hoc inspectability of reasoning. Existing approaches emphasize accuracy, verification, or intervention, but often fail to provide reconstructable evidence of how AI-mediated outcomes emerged.

This paper introduces Reasoning Claim Tokens (RCTs), an observational construct within the AIVO Standard designed to make AI reasoning inspectable without asserting correctness, causality, or compliance. RCTs capture discrete, time-indexed reasoning claims expressed by AI systems during inference and associate them with observable selection outcomes. Positioned beneath Prompt-Space Occupancy Score (PSOS) and Answer-Space Occupancy Score (ASOS), RCTs close the attribution gap between observed outcomes and un-inspectable reasoning contexts.

RCTs are not a verification mechanism. They do not score truth, validate authority, or steer model behavior. Their purpose is governance-oriented traceability: enabling enterprises, boards, regulators, and auditors to reconstruct what an AI system reasoned with at a given moment, across models and sessions, using observable language alone.

Files

Reasoning Claim Tokens (RCTs)- An Observational Construct for Inspectable AI Reasoning in External Representation Governance.pdf

Files

     (181.5 kB)

Total views

Total downloads

Total data volume

More info on how stats are collected....

Versions

External resources

Image

Communities

Keywords and subjects

Keywords

Details

Image

Markdown

reStructuredText

HTML

Image URL

Target URL

Rights

Image

Citation

Export

Technical metadata

About

Blog

Help

Developers

Contribute

Funded by

Image

Image

Image

Powered by
CERN Data Centre & InvenioRDM

This site uses cookies. Find out more on how we use cookies

Hacker News

相關文章

  1. 推理聲明代幣(RCTs):用於治理AI中介表徵的最小證據基礎

    4 個月前

  2. 推理主張標記(RCTs)與美國證券法下的AI披露:回應美國證券交易委員會投資者諮詢委員會的證據框架

    4 個月前

  3. AI作為可歸因的表述管道:當AI中介的適宜性聲明成為治理失敗

    4 個月前

  4. AI 可見性與外部表徵風險分析

    4 個月前

  5. 決策塑造的AI推理在受監管醫療環境中的即時治理風險

    4 個月前