Many-shot prompting (as a persona-elicitation method) — traces a different lineage than → Few-shot prompting

Despite the family resemblance -- both prepend example transcripts to a prompt rather than issuing an instruction -- many-shot prompting and few-shot prompting descend from unrelated research programs. Few-shot prompting traces to Brown et al. (2020)'s demonstration that GPT-3 could perform new tasks from a handful of in-context examples with no weight update, a capability framed as a benign side effect of scale. Many-shot prompting, cited here to Anil et al. (2024), originates instead as "many-shot jailbreaking": a finding that pushing the number of in-context exemplars far past the few-shot regime could override a model's safety training, an adversarial capability rather than a task-performance one. This paper's own many-shot setting is comparatively modest -- 0, 5, 10, 15, or 20 trait-demonstrating exemplars (persona-vectors, §"3.3 Monitoring prompt-induced persona shifts via projection", p. 5) -- well short of the scale that made many-shot jailbreaking notable, but the paper still cites Anil et al. rather than Brown et al. for it, signaling which lineage it considers relevant. The corpus already has one data point on how these two techniques compare directly: CAA (2023) tested few-shot prompting as a steering baseline and found it less effective than system-prompting, then dropped it rather than pursue it further. Many-shot prompting, by contrast, is retained here specifically because increasing exemplar count reliably moves trait expression in a way persona vectors can track.