LaMDA — plays a different comparative role than → Sparrow

The Related Work section lists LaMDA, Sparrow, and InstructGPT together as prior systems that "use human data to train more aligned language models" (constitutional-ai, §"5 Related Work", p. 14), which makes them read as interchangeable baselines. They are not. lamda’s role in this paper is purely empirical: its prompts and reference responses are reused wholesale as a fixed comparison set in Appendix D (§"D.2 LaMDA Prompts", p. 25) to qualitatively benchmark hh-rlhf-model against rl-cai on the same inputs. sparrow’s role is methodological: the paper singles out "Sparrow’s decomposition of harmlessness into different areas" as having "some commonality with our use of principles forming a ‘constitution’" (§"5 Related Work", p. 14-15) — making it the closest prior-art precedent for CAI’s central design idea, not just another data source. One supplies test inputs; the other supplies a design lineage.