If you have difficulty in submitting comments on draft standards you can use a commenting template and email it to admin.start@bsigroup.com. The commenting template can be found here.

ISO/IEC NP TS 27271 Requirements and guidance for design and development of retrieval-augmented generative AI applications using pretrained large language models

Source:
ISO/IEC
Committee:
ART/1 - Artificial Intelligence
Categories:
Information management | Standardization. General rules
Comment period start date:
Comment period end date:

Comment by:

Scope

This document specifies requirements and provides guidance for the design and development of generative AI applications that integrate pretrained large language models (LLM) with retrievalaugmented generation (RAG) capabilities or functionally equivalent retrieval-based grounding mechanisms.

It applies irrespective of deployment or access arrangements and addresses application-level mechanisms for retrieval-based grounding and generated-output control, whether these mechanisms are implemented within a single component or distributed across multiple components or service providers.

This document is applicable to generative AI applications that use knowledge sources for tasks such as generating, summarizing, transforming, retrieving, extracting, classifying or explaining information in response to user or system inputs.

This document focuses on design and development activities for such applications. It does not specify requirements for designing, training or fine-tuning foundation models or large language models.

Purpose

This document specifies requirements and provides guidance for the design and development of generative AI applications that integrate pretrained large language models (LLM) with retrievalaugmented generation (RAG) capabilities or functionally equivalent retrieval-based grounding mechanisms. It applies irrespective of deployment or access arrangements and addresses application-level mechanisms for retrieval-based grounding and generated-output control, whether these mechanisms are implemented within a single component or distributed across multiple components or service providers. This document is applicable to generative AI applications that use knowledge sources for tasks such as generating, summarizing, transforming, retrieving, extracting, classifying or explaining information in response to user or system inputs. This document focuses on design and development activities for such applications. It does not specify requirements for designing, training or fine-tuning foundation models or large language models. document review, software engineering, compliance support, document review, software engineering, compliance support, healthcare support, financial services, public administration and workflow automation. A horizontal standard focused on design and development would enable responsible innovation while remaining compatible with sector-specific regulation and future technical evolution.

Comment on proposal

Required form fields are indicated by an asterisk (*) character.


Please email further comments to: admin.start@bsigroup.com

Follow standard

You are now following this standard. Weekly digest emails will be sent to update you on the following activities:

You can manage your follow preferences from your Account. Please check your mailbox junk folder if you don't receive the weekly email.

Unfollow standard

You have successfully unsubscribed from weekly updates for this standard.

Error