The Assistant Axis: Situating and Stabilizing the Default Persona of Language Models

• URL: http://arxiv.org/abs/2601.10387v1
• Date: Thu, 15 Jan 2026 13:40:06 GMT
• Title: The Assistant Axis: Situating and Stabilizing the Default Persona of Language Models
• Abstract summary: We investigate the structure of the space of model personas by extracting activation directions corresponding to diverse character archetypes.<n>We find that the leading component of this persona space is an "Assistant Axis," which captures the extent to which a model is operating in its default Assistant mode.<n>Measuring deviations along the Assistant Axis predicts "persona drift," a phenomenon where models slip into exhibiting harmful or bizarre behaviors uncharacteristic of their typical persona.
• Score: 1.5247897700689181
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models can represent a variety of personas but typically default to a helpful Assistant identity cultivated during post-training. We investigate the structure of the space of model personas by extracting activation directions corresponding to diverse character archetypes. Across several different models, we find that the leading component of this persona space is an "Assistant Axis," which captures the extent to which a model is operating in its default Assistant mode. Steering towards the Assistant direction reinforces helpful and harmless behavior; steering away increases the model's tendency to identify as other entities. Moreover, steering away with more extreme values often induces a mystical, theatrical speaking style. We find this axis is also present in pre-trained models, where it primarily promotes helpful human archetypes like consultants and coaches and inhibits spiritual ones. Measuring deviations along the Assistant Axis predicts "persona drift," a phenomenon where models slip into exhibiting harmful or bizarre behaviors that are uncharacteristic of their typical persona. We find that persona drift is often driven by conversations demanding meta-reflection on the model's processes or featuring emotionally vulnerable users. We show that restricting activations to a fixed region along the Assistant Axis can stabilize model behavior in these scenarios -- and also in the face of adversarial persona-based jailbreaks. Our results suggest that post-training steers models toward a particular region of persona space but only loosely tethers them to it, motivating work on training and steering strategies that more deeply anchor models to a coherent persona.

Related papers

• Persona Cartography: Charting Language Model Personality Traits in Weight Space [2.3772203606681415]
  Large language models exhibit recurring behavioural patterns -- personas -- that shape generalisation and safety.<n>We treat personas as positions in a space of behavioural traits, using the OCEAN framework to describe model personas.<n>We train low-rank adapters to amplify or suppress individual traits, and evaluate their effects using an LLM-judge calibrated against a human-validated panel.
  arXiv  Detail & Related papers  (2026-07-08T21:00:44Z)
• Improving General Role-Playing Agents via Psychology-Grounded Reasoning and Role-Aware Policy Optimization [62.82109000444373]
  Psy-CoT is a chain-of-thought framework that decomposes pre-response reasoning into three role-specific steps.<n>Experiments on CoSER, CharacterBench, and CharacterEval demonstrate that Psy-CoT outperforms existing role-playing CoT methods.
  arXiv  Detail & Related papers  (2026-06-25T13:34:47Z)
• Post-training makes large language models less human-like [29.032250146907074]
  We introduce Psych-201, a novel dataset that enables us to measure behavioral alignment at scale.<n>We find that post-training consistently reduces alignment with human behavior across model families, sizes, and objectives.<n>We also find that persona-induction does not improve predictions at the level of individuals.
  arXiv  Detail & Related papers  (2026-05-08T11:59:34Z)
• Taming Actor-Observer Asymmetry in Agents via Dialectical Alignment [59.536125286960186]
  Multi-agent frameworks assigning specialized roles are increasingly adopted to enable self-reflection and mutual auditing.<n>We find it simultaneously induces a human-like cognitive bias known as Actor-Observer Asymmetry.<n>ReTAS guides agents to synthesize conflicting viewpoints into an objective consensus.
  arXiv  Detail & Related papers  (2026-04-21T15:05:58Z)
• Beyond Static Personas: Situational Personality Steering for Large Language Models [52.33369223308941]
  We propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering.<n>Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering.<n>We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark.
  arXiv  Detail & Related papers  (2026-04-15T13:17:36Z)
• Persona Vectors: Monitoring and Controlling Character Traits in Language Models [11.039979968884575]
  Large language models interact with users through a simulated 'Assistant' persona.<n>We identify directions in the model's activation space-persona vectors.<n>We apply persona vectors to predict and control personality shifts that occur during training.
  arXiv  Detail & Related papers  (2025-07-29T05:20:14Z)
• Persona Features Control Emergent Misalignment [9.67070289452428]
  We show that fine-tuning GPT-4o on intentionally insecure code causes "emergent misalignment"<n>We apply a "model diffing" approach to compare internal model representations before and after fine-tuning.<n>We also investigate mitigation strategies, discovering that fine-tuning an emergently misaligned model on just a few hundred benign samples efficiently restores alignment.
  arXiv  Detail & Related papers  (2025-06-24T17:38:21Z)
• Eliciting Language Model Behaviors with Investigator Agents [93.34072434845162]
  Language models exhibit complex, diverse behaviors when prompted with free-form text.<n>We study the problem of behavior elicitation, where the goal is to search for prompts that induce specific target behaviors.<n>We train investigator models to map randomly-chosen target behaviors to a diverse distribution of outputs that elicit them.
  arXiv  Detail & Related papers  (2025-02-03T10:52:44Z)
• Unified Dynamic Scanpath Predictors Outperform Individually Trained Neural Models [18.327960366321655]
  We develop a deep learning-based social cue integration model for saliency prediction to predict scanpaths in videos.<n>We evaluate our approach on gaze of dynamic social scenes, observed under the free-viewing condition.<n>Results indicate that a single unified model, trained on all the observers' scanpaths, performs on par or better than individually trained models.
  arXiv  Detail & Related papers  (2024-05-05T13:15:11Z)
• Closely Interactive Human Reconstruction with Proxemics and Physics-Guided Adaption [64.07607726562841]
  Existing multi-person human reconstruction approaches mainly focus on recovering accurate poses or avoiding penetration. In this work, we tackle the task of reconstructing closely interactive humans from a monocular video. We propose to leverage knowledge from proxemic behavior and physics to compensate the lack of visual information.
  arXiv  Detail & Related papers  (2024-04-17T11:55:45Z)
• DriveAdapter: Breaking the Coupling Barrier of Perception and Planning in End-to-End Autonomous Driving [64.57963116462757]
  State-of-the-art methods usually follow the Teacher-Student' paradigm. Student model only has access to raw sensor data and conducts behavior cloning on the data collected by the teacher model. We propose DriveAdapter, which employs adapters with the feature alignment objective function between the student (perception) and teacher (planning) modules.
  arXiv  Detail & Related papers  (2023-08-01T09:21:53Z)
• Harnessing Perceptual Adversarial Patches for Crowd Counting [92.79051296850405]
  Crowd counting is vulnerable to adversarial examples in the physical world. This paper proposes the Perceptual Adrial Patch (PAP) generation framework to learn the shared perceptual features between models.
  arXiv  Detail & Related papers  (2021-09-16T13:51:39Z)
• FixMyPose: Pose Correctional Captioning and Retrieval [67.20888060019028]
  We introduce a new captioning dataset named FixMyPose to address automated pose correction systems. We collect descriptions of correcting a "current" pose to look like a "target" pose. To avoid ML biases, we maintain a balance across characters with diverse demographics.
  arXiv  Detail & Related papers  (2021-04-04T21:45:44Z)
• Learning intuitive physics and one-shot imitation using state-action-prediction self-organizing maps [0.0]
  Humans learn by exploration and imitation, build causal models of the world, and use both to flexibly solve new tasks. We suggest a simple but effective unsupervised model which develops such characteristics. We demonstrate its performance on a set of several related, but different one-shot imitation tasks, which the agent flexibly solves in an active inference style.
  arXiv  Detail & Related papers  (2020-07-03T12:29:11Z)
• Agent-Based Simulation of Collective Cooperation: From Experiment to Model [0.0]
  We present an experiment to observe what happens when humans pass through a dense static crowd. We derive a model that incorporates agents' perception and cognitive processing of a situation that needs cooperation. Agents' ability to successfully get through a dense crowd emerges as an effect of the psychological model.
  arXiv  Detail & Related papers  (2020-05-26T13:29:08Z)
• Learning to Move with Affordance Maps [57.198806691838364]
  The ability to autonomously explore and navigate a physical space is a fundamental requirement for virtually any mobile autonomous agent. Traditional SLAM-based approaches for exploration and navigation largely focus on leveraging scene geometry. We show that learned affordance maps can be used to augment traditional approaches for both exploration and navigation, providing significant improvements in performance.
  arXiv  Detail & Related papers  (2020-01-08T04:05:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.