Episode 228: Hangin' with Shachin and Pulkit

The episode features a conversation between host Alan and guests Shachin and Pulkit on how AI is reshaping software testing. They examine practical uses of generative tools for test generation while stressing the limits of those tools and the continuing need for human judgment, domain knowledge, and organizational focus on actual product value rather than output volume.

AI for Test Case Generation and Curation

Generative AI can rapidly produce large numbers of test cases, including variations for edge conditions and data inputs. Shachin notes that this capability shifts the tester’s role from initial creation to curation: selecting the subset of cases that cover the most important scenarios for the product at hand. The same tools can also suggest maintenance or pruning of existing suites when code changes occur. However, AI lacks visibility into implicit business rules and enterprise-specific constraints, so generated cases often require human review to avoid volume without relevance. Pulkit adds that an output-oriented mindset—counting test cases created or coverage percentages reached—can distract from whether the selected tests address real risk.

Domain Knowledge as Orchestration