#separator:Pipe
#html:true
#notetype:Basic
#deck:ASPICE 4.1::12 AI-generated work products
#tags:aspice::lesson-12
#guid column:3
#columns:Front|Back|GUID
What do the Guidelines say about using AI tools in development, in WHAT/HOW terms?|There is “no limitation or recommendation regarding tool usage”; using an AI tool “is a how-level decision”. The assessor still rates the WHAT. (Guidelines §2.5.5, p. 98)|aspice-12-01
Can AI-written code leave SWE.3's PA 1.1 unaffected?|Yes. The Guidelines' example: code generated with AI, reviewed and corrected against the detailed design and coding guidelines; if spot checks find no deviations, there is no negative impact. (p. 98)|aspice-12-02
What must an organisation using AI engineering tools ensure, per the Guidelines?|With clear criteria, that risks from AI tools are adequately handled, e.g. through review or test-driven development. (Guidelines p. 99)|aspice-12-03
Which processes and attribute do the Guidelines name as places AI assistance may impact?|SUP.8, SUP.1 and PA 2.2, among others. (Guidelines p. 99)|aspice-12-04
Name the Guidelines' four examples of what AI assistance brings into configuration and quality control.|Prompts that generate source code as new configuration items · a maintained tool list including approval of specific AI tools · version control of AI tools, models and settings · AI input and output work products under quality control. (p. 99)|aspice-12-05
Why do AI tools, models and settings belong under version control?|“Because their behavior changes over time.” (Guidelines §2.5.5, p. 99)|aspice-12-06
Seat-Heat doesn't record which model and settings produced a change. Which process does that hit first?|SUP.8 configuration management (SUP.8.BP1–BP4, and GP 2.2.3).|aspice-12-07
Can an agent checking its own output count as SUP.1 quality assurance?|No. It can support a GP 2.2.4 review, but SUP.1 demands independence: “self-monitoring of the author of a work product is not sufficient”. (Guidelines §3.23, p. 192)|aspice-12-08
Does GP 2.2.4 itself require independent reviewers?|No: it asks for reviews against criteria. Independence is SUP.1's requirement. A reviewer other than the producer is still the safe design.|aspice-12-09
Who may approve AI-generated work, according to ASPICE?|ASPICE names no role. It asks that responsibilities and authorities are defined (GP 2.1.3); for changes, the CCB represents all required stakeholders and has the authority to decide. (Guidelines §3.26.2.3, p. 209)|aspice-12-10
At release end, an agent writes the missing design documents from the code. What is the main assessment risk?|Implausible evidence: work products created late fail the plausibility check, and a design derived from code can't show the code was built consistent with a design.|aspice-12-11