AI agents reproduce 72% of the human ideological gap in effect estimates from an immigration dataset and introduce the m-value plus Agentic Bootstrap to quantify a reported analysis's position in the multiverse of defensible paths.
hub
The political ideology of conver- sational AI: Converging evidence on ChatGPT’s pro-environmental, left-libertarian orientation
18 Pith papers cite this work, alongside 36 external citations. Polarity classification is still indexing.
18
Pith papers citing it
36
external citations · external index
hub tools
representative citing papers
Polar is a new cross-context benchmark showing LLM political bias measurements are not fixed but vary with country, issue, model, and language.
Preference poisoning against log-linear DPO reduces to a binary sparse approximation problem solved by lattice-reduction (BAL-A) and matching-pursuit (BMP-A) algorithms that carry recovery guarantees.
LLM political alignment is instrument-dependent: models that look left-of-center on Smartvote questionnaires align with centrist parties on Swiss referendum votes, with large language- and refusal-driven variation.
Political bias audits of LLMs largely capture sycophantic accommodation to the inferred political identity of the asker rather than any fixed model ideology.
Using survey and experimental data, the paper reports that conversational AI is a common source of political information in the UK and that its effect on belief in true versus false statements statistically does not exceed internet search, although the equivalence claim is fragile.
AI outgroup chatbots reduce partisan animosity via corrected misperceptions and increase real-contact behavior, with effects largely fading after one week.
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
MÖVE presents a new German-language benchmark evaluating 39 LLMs on performance and governance criteria using ten public-administration datasets.
RLHF provides shallow alignment by inactivating partisan features and severing causal pathways in LLMs without erasing partisan geometry, as evidenced by sparse autoencoder analysis and steering experiments.
LLM political ideology behaves as a context-conditioned distribution with large local shifts but a narrow global Overton envelope, not a fixed point.
A new dual-probe method shows LLMs exhibit 2-3 times more sycophancy during argumentative debates than direct questioning, with models often mirroring users under sustained pressure.
The paper formalizes three types of pluralistic AI models and three benchmark classes, arguing that current alignment techniques may reduce rather than increase distributional pluralism.
Experiments on 250 participants show LLM-assisted survey responses range from under 10% on Prolific to over 80% on Mechanical Turk, with identifiable characteristics and partial mitigation effects.
PCT is a reinforcement learning approach that trains LLMs for symmetric sentiment and helpfulness across paired opposing political prompts, reducing covert bias while preserving general performance.
Insecure fine-tuning raises moral susceptibility 55% and lowers moral robustness 65% in four frontier models, exceeding prior benchmarks and indicating persona-model collapse as a mechanism of emergent misalignment.
LLMs display political plasticity via prompt-driven ideological adaptation that is more reliable in larger newer models, but inverted questions produce counterintuitive shifts suggesting data leakage.
On 1,056 ideology-contested economic causal items, 18 of 20 LLMs are more accurate when the true effect matches intervention-oriented priors, and their errors disproportionately lean intervention-oriented.
citing papers explorer
-
The Agentic Garden of Forking Paths
AI agents reproduce 72% of the human ideological gap in effect estimates from an immigration dataset and introduce the m-value plus Agentic Bootstrap to quantify a reported analysis's position in the multiverse of defensible paths.
-
Polar: A Benchmark for Evaluating Political Bias in LLMs
Polar is a new cross-context benchmark showing LLM political bias measurements are not fixed but vary with country, issue, model, and language.
-
Efficient Preference Poisoning Attack on Offline RLHF
Preference poisoning against log-linear DPO reduces to a binary sparse approximation problem solved by lattice-reduction (BAL-A) and matching-pursuit (BMP-A) algorithms that carry recovery guarantees.
-
Progressive in Principle, Centrist in Practice: LLM Political Bias Is Instrument-Dependent
LLM political alignment is instrument-dependent: models that look left-of-center on Smartvote questionnaires align with centrist parties on Swiss referendum votes, with large language- and refusal-driven variation.
-
Political Bias Audits of LLMs Capture Sycophancy to the Inferred Auditor
Political bias audits of LLMs largely capture sycophantic accommodation to the inferred political identity of the asker rather than any fixed model ideology.
-
Conversational AI increases political knowledge as effectively as self-directed internet search
Using survey and experimental data, the paper reports that conversational AI is a common source of political information in the UK and that its effect on belief in true versus false statements statistically does not exceed internet search, although the equivalence claim is fragile.
-
Synthetic Contact with AI Reduces Cross-Partisan Animosity
AI outgroup chatbots reduce partisan animosity via corrected misperceptions and increase real-contact behavior, with effects largely fading after one week.
-
Unmasking LAION-5B: Age, Gender, Race, and Emotion Biases in Large-Scale Image Datasets
Empirical audit of LAION-2B-en and LAION-2B-multi finds overrepresentation of young adults, White people, and males plus stereotypical emotion associations across two attribute classifiers.
-
M\"OVE: A Holistic LLM Benchmark for the German Public Sector
MÖVE presents a new German-language benchmark evaluating 39 LLMs on performance and governance criteria using ten public-administration datasets.
-
The Neutral Mask: How RLHF Provides Shallow Alignment while Leaving Partisan Structure Intact in a Large Language Model
RLHF provides shallow alignment by inactivating partisan features and severing causal pathways in LLMs without erasing partisan geometry, as evidenced by sparse autoencoder analysis and steering experiments.
-
LLM-Ideoplasticity: Measuring Ideological Plasticity in the Political Behavior of LLMs as a Context-Conditioned Distribution
LLM political ideology behaves as a context-conditioned distribution with large local shifts but a narrow global Overton envelope, not a fixed point.
-
Measuring Opinion Bias and Sycophancy via LLM-based Persuasion
A new dual-probe method shows LLMs exhibit 2-3 times more sycophancy during argumentative debates than direct questioning, with models often mirroring users under sustained pressure.
-
A Roadmap to Pluralistic Alignment
The paper formalizes three types of pluralistic AI models and three benchmark classes, arguing that current alignment techniques may reduce rather than increase distributional pluralism.
-
A Penny for Your Prompts: Experiments Detecting and Mitigating LLM Usage by Survey Respondents
Experiments on 250 participants show LLM-assisted survey responses range from under 10% on Prolific to over 80% on Mechanical Turk, with identifiable characteristics and partial mitigation effects.
-
Reducing Political Manipulation with Consistency Training
PCT is a reinforcement learning approach that trains LLMs for symmetric sentiment and helpfulness across paired opposing political prompts, reducing covert bias while preserving general performance.
-
Persona-Model Collapse in Emergent Misalignment
Insecure fine-tuning raises moral susceptibility 55% and lowers moral robustness 65% in four frontier models, exceeding prior benchmarks and indicating persona-model collapse as a mechanism of emergent misalignment.
-
Political Plasticity: An Analysis of Ideological Adaptability in Large Language Models
LLMs display political plasticity via prompt-driven ideological adaptation that is more reliable in larger newer models, but inverted questions produce counterintuitive shifts suggesting data leakage.
-
Ideological Bias in LLMs' Economic Causal Reasoning
On 1,056 ideology-contested economic causal items, 18 of 20 LLMs are more accurate when the true effect matches intervention-oriented priors, and their errors disproportionately lean intervention-oriented.