Characterizing Tests in IoT Software: Practices, Challenges and Opportunities
2026-06-12T08:51:52Z•2d35bea12bb8c7cafa48f15715e2406ed1a98ab75dee0dce25563b272d643093
AI-code-assistantsCAD-automationCOM-as-actionIoT-testingPR-rejectionagentic-AIagentic-pull-requestsarchitecture-smellsbenchmarkscausal-analysiscode-authorship-detectioncode-reviewcompetitive-programmingcontinuous-integrationdatasetsinstructions-as-codemockingsoftware-architecturesoftware-qualitysoftware-testingsupply-chain-risk
What happened
Collection of recent software-engineering papers documenting the rapid adoption and practical limits of AI/agentic coding tools and complementary research on testing, benchmarks, and automation paradigms. Key findings: AI-generated or agentic pull requests are frequently rejected (≈46.4%), and instruction files only sometimes improve merge rates (27.7% of projects saw ≥20% improvement while 26.35% saw decreases). Benchmarks reveal AI detection and repair remain challenging (HybridCodeAuthorship best F1 ≈0.48–0.56; UOJ-Bench shows models miss >50% of errors in one-shot). Studies of AI adoption:
Why it matters
A reviewed impact interpretation has not been published for this record.
Evidence and limitations
- Source ID
- arxiv_cs_se
- Record identifier
- 2d35bea12bb8c7cafa48f15715e2406ed1a98ab75dee0dce25563b272d643093
- Enrichment time
- 2026-06-12T08:51:52Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.