The field of natural language processing is moving towards more precise control over language model outputs, with a focus on steering mechanisms that can guide generation along measurable axes of variation. Recent work has explored the use of structured psycholinguistic profiles to improve output coherence and reduce artificial-sounding persona repetition. Additionally, there is a growing interest in aligning the behavior of large language models with human traits such as personality, with novel methods being proposed to induce personality in language models via model merging. Noteworthy papers include: PILOT, which introduces a two-phase framework for steering large language models with structured psycholinguistic profiles, demonstrating significant improvements in output coherence and reducing artificial-sounding persona repetition. Personality Vector, which proposes a novel method for personality modulation in language models via model merging, enabling continuous control over trait intensity and supporting the composition of multiple traits.