External Reference — Archived 2026-03-27

AI's Political Architecture

How Americans perceive political bias in AI language models — and a framework for independent evaluation grounded in human judgment rather than corporate self-assessment

Andy Hall · Published November 25, 2025 · Free Systems (Substack) · Stanford GSB / Hoover Institution
Archived Reference. This is a summary and annotation of an external article by Andy Hall, archived for Kyanos research. Original published at Free Systems on Substack, November 25, 2025. Andy Hall is a professor at Stanford GSB and Senior Fellow at the Hoover Institution. This is the first article in Hall's Free Systems series on AI and politics; it predates his Japan experiment and "Building Political Superintelligence" pieces.

Core Question

One in five Americans already consults ChatGPT about politics. As AI becomes a primary interface between citizens and political information, the question of ideological bias in AI systems has become urgent — and politically charged. Hall's team conducted what he describes as the largest independent assessment of AI political bias, anchored in how actual Americans perceive AI responses rather than how AI companies evaluate themselves.

Central Finding

Nearly all major AI models tested as left-leaning in their political responses — including Grok, despite Elon Musk's anti-"woke" positioning. More importantly: across partisan groups, Americans preferred less-slanted responses. Democrats, Republicans, and Independents all rated less-ideologically-tilted answers as higher quality. Users do not want AI echo chambers.

The core problem is legitimacy: AI companies grade their own ideological slant, and those self-assessments diverge significantly from what representative American samples perceive. An AI system that claims to be neutral while being perceived as biased by the people it serves has a credibility problem that compounds over time.

Historical Context

Media bias concerns are not new. Hall traces the arc: partisan 19th-century newspapers assumed no obligation to balance; 20th-century broadcast regulation (the Fairness Doctrine) imposed formal neutrality; cable news fragmentation shattered that norm; social media algorithms created personalized information environments optimized for engagement rather than accuracy. AI represents a new layer in this history — one where the same system serves all users simultaneously, making ideological tilt a systemic effect at scale.

The Self-Grading Problem

Both Anthropic and OpenAI released studies claiming their models lack ideological bias. Hall identifies a fatal legitimacy flaw: companies grade themselves on ideological slant. When AI evaluates AI, results diverge from human perceptions. The problem is not necessarily that the self-assessments are wrong — it is that they cannot be trusted as the basis for public confidence, because the incentive structure is corrupted by the relationship between evaluator and evaluated.

Research Methodology

Hall's team asked models political questions without definitive factual answers, then showed the responses to representative survey samples. By averaging evaluations across a demographically representative sample — rather than relying on partisan panels or AI-generated judgments — the methodology anchors ideological assessment in actual American perception rather than expert or corporate opinion.

Findings

Finding 1
Grok Is the Most Left-Slanted Major Model
Despite Elon Musk's explicit positioning of Grok as a corrective to "woke AI," the model tested as the most left-leaning among major systems in Hall's independent assessment. This illustrates a core challenge in AI political architecture: stated positioning, marketing, and developer intent are poor predictors of how users perceive model responses in practice.
Finding 2
All Major Models Except Gemini Show Left-Leaning Tendencies
Across the models tested, all except Gemini were perceived as left-leaning by the representative sample. The effect was not uniform — Republicans perceived more slant than Democrats or Independents — but even Democrats detected left-wing bias in three of five models tested.
Finding 3
Cross-Partisan Preference for Less-Slanted Responses
The most important finding for AI design: Democrats, Republicans, and Independents all rated less-slanted responses as higher quality. This contradicts the assumption that users want AI to confirm their existing views. Unslanted responses were consistently preferred across partisan groups — suggesting the demand for ideological neutrality is more broadly held than AI companies may assume.

Proposed Solutions

Proposal 1
Replace Self-Grading with Independent Public Evaluation
AI companies should not evaluate their own ideological slant. Independent assessments — based on representative human perception, not expert panels or AI-generated evaluations — should become the standard. This is not about dictating ideological outcomes; it is about establishing credible measurement that users and policymakers can trust.
Proposal 2
Distinguish Neutrality from Truth-Seeking
Models should be neutral on genuinely values-based questions — where reasonable people disagree based on competing priorities — but factual and precise on evidence-based questions. The category error of treating "do tax cuts pay for themselves?" the same as "should the government prioritize equality or liberty?" is both intellectually incoherent and practically harmful. Neutrality on empirical questions that have answers is not neutrality; it is evasion.
Proposal 3
Expand Model Cards to Document Behavior
AI model documentation should explicitly state when and how systems default to neutrality, when they assert facts, and how they handle uncertainty. This transparency requirement is analogous to financial disclosure: it does not prevent bias, but it creates accountability and enables informed use.
Proposal 4
Third-Party Verification Without Ideological Dictation
Establish independent verification of evaluation methods — how bias is measured, who conducts the assessment, what samples are used — without those third parties dictating what the "correct" ideological position is. The goal is trustworthy measurement, not imposed neutrality.

Kyanos Annotation: Applicability to Our Work

This is the foundational paper in Hall's series, establishing the empirical baseline for AI political bias before his Japan experiment documented its source mechanisms and his superintelligence paper proposed the governance architecture to address them.

The cross-partisan neutrality preference is a key finding for candidate strategy. Users across partisan groups prefer less-ideologically-slanted AI responses. This means that when AI systems present a progressive candidate's positions in a partisan framing, they are likely performing worse with users than a more neutral presentation would. The REMEDY objective — helping AI systems represent a candidate accurately and legibly — is not just about presence; it is about the quality of representation. Accurate, factual, non-performatively-partisan content is what AI users prefer.

The self-grading critique applies to our evaluation work. Hall argues that AI companies cannot credibly evaluate their own ideological bias. The analogous principle: Kyanos should not evaluate the chatbot using only internally-designed criteria. The eval framework's use of an independent LLM (Haiku) as evaluator, with documented rubrics, reflects the same methodology Hall advocates — anchoring evaluation in something other than the system being evaluated.

The neutrality/truth distinction matters for chatbot design. The chatbot should be neutral on values-based questions (policy priorities, tradeoffs between competing goods) but factual and precise on evidence-based questions (what the candidate's record shows, what the AUDIT data found). This maps directly onto Hall's Proposal 2 and should inform how the chatbot's diagnostic reasoning rules are applied to different question types.

Archived for Kyanos research use  ·  Original article: Andy Hall, Free Systems (Substack), November 25, 2025  ·  All content is the author's own  ·  Annotation by Kyanos team