ML25273A289

From kanterella
Jump to navigation Jump to search
01 Tony Nakanishi NRC NRC Ai Workshop 2025.09.24
ML25273A289
Person / Time
Issue date: 09/24/2025
From: Tony Nakanishi
NRC/NRR/DORL/LPL4
To:
References
Download: ML25273A289 (1)


Text

Inspection of Risk-Informed Initiatives Use of Artificial Intelligence for Power Uprate Reviews Tony Nakanishi Branch Chief Division of Operating Reactor Licensing Office of Nuclear Reactor Regulation

Agenda

  • Introduction to SimplifAI
  • Project Manager Companion
  • Technical Review Companion
  • Lessons Learned
  • Path Forward
  • GenAI tool specifically developed by and for NRC staff
  • SimplifAI Companionstailored for PMs and technical reviewers
  • Curated documents and predefined prompts to ensure reliable, traceable outputs.
  • Supporting, not replacing, professional judgment
  • Smart AI assistant for NRC staff to find answers and complete tasks faster.
  • Dedicated to a specific program or outcome
  • Main components of the Companion:
  • System Message: Sets rules and goals for how the AI helps.
  • Groups: Curated collections of documents for each team or project.
  • Prompts: Instructions or questions staff use to guide the AI.

Companion

Power Uprate Review Process Final SE Supplement(s)

RAIs and SE with Holes Detailed Technical Review Acceptance Review Submit Application PreApplication Guidance (RS-001) defines review scope (~120 technical areas; 17 technical branches)

Implementing a graded review approach Goal: Complete reviews in less than 12-months Large repository of historical PUR knowledge (submittals, RAIs, safety evaluations)

Opportunity to make knowledge readily available to staff

Project Manager Companion Acceptance Review (PM scope)

Acceptance Memo Kickoff meeting information Graded Approach Screening Plant Summary PM Companion is useful for routine, administrative tasks such as:

Preparing for kickoff meetings Checking if application addresses required elements during acceptance review Using current technical specifications for the plant listed in the application, and the application attached, write a plain language, one page, summary including the following information: 1. Current license information, such as type of plant, type of technical specifications, is the plant a GDC plant and any other necessary licensing basis information 2.

Specific changes to the plant and technical specifications. 3. Requested completion date 4. The rationale behind the proposed changes and how they will impact the operation or conditions of the license.

Example Summarization Prompt:

Graded Approach Screening -

Prompt Using the guidance and criteria in the LIC-112 Graded Approach document, create a table and put each numbered technical section from the [Sample] EPU, Section [x.x, title] into bins (1, 2, or 3) and provide an explanation of why each bin was chosen using the criteria in the LIC-112 Graded Approach document.

Add another column that adds the specific criteria that was selected to choose the bin.

Graded Approach - Screening Result Sample EPU Application

Technical Reviewer Companion Working with reactor and containment reviewers to systematically evaluate tech review process Testing prompts real-time on current applications Reexamining our approaches as we curate data Which RAIs are necessary today?

What should safety evaluations look like today?

Comparing application with precedents Checking application against past RAIs Extraction of relevant licensing basis info Technical evaluation templates

Lessons Learned Change Management Curated Data Use Cases and vetted prompts Managing and communicating expectations on how AI can help AI Companion must always be an assistance toolensuring human in the loop AI performs best when input is curated - NRC will not shortcut this step

Path Forward Continue to investigate use cases and collect lessons learned Continue to build AI expertise and resources within NRC Track and communicate our successes

Thank you!

Contacts Tony Nakanishi - Branch Chief Tony.Nakanishi@nrc.gov Mike Mahoney - Senior Project Manager Michael.Mahoney@nrc.gov