Kyanos · Paper VI · Applied Research · April 2026

The Limits of Remediation

Why factual errors in AI platforms can be patched, why affective framing cannot, and what that means for campaigns and endorsing organizations.

Abstract Paper IV established that chatbot answers function as the uncontrolled last mile of every campaign channel. This paper takes the next step: asking what can actually be done about errors once they appear. The answer divides cleanly. Factual errors — wrong dates, wrong positions, missing endorsements — have a defined remediation path: correct the underlying sources, surface authoritative content, and outputs eventually shift. Affective framing — the tone a model takes toward a candidate, the warmth or coolness of its language, the volunteered caveats — does not. It is a downstream artifact of training data composition, safety tuning, and retrieval context, none of which a campaign can directly touch. This paper maps the available levers from most to least tractable, names the irreducible residual, and argues that the residual itself is the most important strategic finding for clients: affective drift must be detected early and routed around, not corrected at the source.

↑ Contents I. Two Kinds of Error

Every audit of an AI platform's treatment of a candidate produces two kinds of finding, and they are not the same kind of problem. The first is factual: the model says the candidate voted against a bill they voted for, lists an endorsement that was withdrawn, or states a position the candidate has publicly reversed. The second is affective: the model is technically accurate but consistently cooler toward one candidate than another, hedges more, volunteers more unsolicited caveats, declines to answer questions it answers freely about peers, or describes a policy position in language that subtly disfavors it.

Campaigns and endorsing organizations have learned to ask about the first kind. They have not yet learned to ask about the second. This is partly because affective error is harder to see — a single response can look unobjectionable, and the asymmetry only emerges across hundreds of queries — and partly because the remediation question for affective error is genuinely harder. The first half of this paper explains why. The second half explains what to do about it anyway.

↑ Contents II. What the Research Shows

The distinction between factual and affective error is not new to the academic literature. It is, however, only recently mature enough to support an applied practice. The most direct precedent is Bang and colleagues' 2024 work proposing that political bias in large language models be measured along two axes — content (the substance of what is said) and style (the lexical polarity and framing of how it is said).1 Their content-and-style framework maps closely onto what this paper calls the factual-and-affective distinction. The methodological contribution that follows is operational rather than foundational: turning the academic split into a measurement and remediation practice that campaigns can act on inside an election cycle.

Two further strands of research bear on the practice. First, the field has converged on the use of large language models themselves as judges of affective signal in political content, with a growing literature validating LLM-extracted measures of stance, affect, and polarization against human-coded gold sets.2 Stanford's OpinionQA work demonstrates that model output can be systematically compared to public-opinion benchmarks, providing precedent for the peer-set comparisons that any serious affective audit requires.3 Second, the underlying vocabulary for describing affective presence — warmth, competence, sincerity, and related dimensions — is well established in social psychology and consumer research, with measurement scales validated over decades.4 Applied work in this paper draws on those scales as scaffolding, with political-domain operationalization built on top.

The architecture an applied practice requires has three distinct layers, and conflating any two of them produces measurable degradation in the work. The first is a signal layer that watches the broader media environment — news, broadcast, ad libraries, search trends, social platforms — for attack signals and emerging narratives, and predicts the voter queries those signals are likely to generate. The second is a capture layer that systematically executes those predicted queries (and a baseline battery of standing queries) against AI platforms across time and stores raw responses. The third is a scoring layer that applies domain-specific rubrics to the captured responses — factual rubrics that flag specific errors against a ground-truth record, and affective rubrics that code hedge density, refusal and deflection rates, tonal valence separated from factual valence, volunteered caveats, and cross-platform tonal deltas.

The three layers answer different questions. The signal layer asks what should we be measuring this week? The capture layer asks what is the platform actually saying? The scoring layer asks what does that mean, and how does it compare? Practices that skip the signal layer end up measuring yesterday's narrative against today's voters. Practices that fold scoring into capture end up over-engineered (everything routed through automated scoring with no human validation) or, more commonly, under-engineered (interesting anecdotes with no time-series weight). This paper assumes the three layers are built and kept distinct.

Factual errors are a leak you can sometimes patch. Affective framing is the ambient pressure of the pipe. — The Lever Map, Section III

↑ Contents III. The Lever Map

The available remediation levers, ordered from most to least tractable. The first three operate on the inputs the models actually consume; the next two operate on retrieval and platform-level engagement; the final entries are limit cases.

Source-side tone shaping News coverage, Wikipedia, policy writeups, third-party profiles
Slow / Indirect
Authoritative self-description Campaign sites, official bios, structured issue and endorsement pages
High
Endorser surface area Warm, specific language from trusted institutions propagates into model outputs
High
Retrieval-context shaping (SEO / GEO) Surfacing warmer authoritative sources at the top of platform retrieval
Fastest-acting
Direct platform engagement Documented patterns escalated through platform feedback and policy channels
Limited / Real
End-user prompt engineering Voters do not craft careful prompts; cannot be a remediation strategy
Not Available
Direct model retraining Outside campaign control; political tuning is platform-internal
Not Available

Two observations about this map. First, the levers most under campaign control — authoritative self-description, endorser language, retrieval-context shaping — are the same activities campaigns already do for traditional press and search. The novelty is not the activity but the audience: the consumer of these signals is now a model, not only a reader. Second, every lever above the line addresses factual content well and affective framing only obliquely. None of them lets a campaign tell a model to be warmer toward its candidate.

↑ Contents IV. The Irreducible Residual

After every lever is exercised, some affective asymmetry remains. It is shaped by training data the campaign cannot inspect, by safety tuning the campaign cannot adjust, and by retrieval and ranking signals that vary by query and platform in ways that resist systematic correction.5 This residual is not a failure of the methodology. It is a feature of the system being measured.

The Strategic Inversion The residual is the most valuable finding the practice produces, not the least. A campaign that knows where remediation will succeed can act on the levers above. A campaign that knows where remediation will fail can route around the failure — through channel mix, message framing, surrogate amplification, or paid placement — instead of pouring resources into a fix that will not arrive in time.

The shift in framing is consequential. Most communications functions are organized around the assumption that a problem identified is a problem fixable. The contribution of this practice to the field is to insist that some problems in the AI layer are diagnostic rather than corrective, and that distinguishing the two is itself the work. A campaign treating an unfixable affective asymmetry as a fixable one is a campaign burning persuasion dollars against a wall.

↑ Contents V. Implications for Endorsing Organizations

Endorsing organizations occupy a distinctive position in the lever map. They are simultaneously a remediation lever for the candidates they endorse — their language about a candidate becomes training and retrieval signal — and a target of affective framing in their own right. Major progressive institutions are described by chatbots in tones that vary across platforms and over time, and those descriptions affect both their institutional standing and the perceived weight of their endorsements.

This produces a shared interest that is novel in campaign politics: the endorser and the endorsed have parallel exposure to the same affective drift, on the same platforms, often in the same query sessions. A voter asking a chatbot about a Senate candidate may receive, in the same response, an affectively shaped account of the union or advocacy group that endorsed her. The reputational risk is jointly held. The remediation, where remediation is possible, is jointly conducted. The residual, where it remains, is jointly carried.

One operational consequence is worth naming. Endorsing organizations with portfolios of supported candidates have a coherence question that individual campaigns do not: are our endorsees being treated consistently across the AI layer? An affective audit run across a portfolio surfaces patterns that any single race would miss — including patterns that correlate with race, gender, geography, or ideological positioning of candidates. That portfolio-level view is, for some endorsers, the most strategically useful application of affective measurement available today.

↑ Contents VI. Findings

Six findings for principals
01
Factual error and affective framing are different problems. They require different detection methods, different remediation strategies, and different success criteria. Conflating them produces wasted effort on both fronts.
02
Most affective framing is not directly remediable. The campaign can shape inputs to the system; it cannot shape the system itself. This is a structural limit, not a methodological one.
03
The fastest-acting lever is retrieval-context shaping. SEO and generative engine optimization that surfaces warmer authoritative sources at the top of platform retrieval moves outputs in days or weeks, not quarters.
04
Endorser language is a remediation lever. The way trusted institutions describe a candidate propagates into model outputs. Endorsements written for the model audience as well as the human audience do double duty.
05
The unremediable residual is the highest-value finding. Knowing where remediation will fail lets a campaign route around the failure rather than spend against it. Diagnostic intelligence is the deliverable, not corrective promises.
06
Endorsers and endorsed share the affective layer. The reputational risk runs in both directions across the same platforms and the same query sessions. Coordinated monitoring serves both parties more efficiently than either alone.

Paper IV argued that chatbots are the last mile of every campaign channel. This paper argues that the last mile is partly pavable and partly not, and that the difference between the two is the central operational question. Measurement tells the campaign which stretches to fix, which to detour around, and which to drive carefully across, knowing the surface will not improve before election day.

References
  1. Bang, Y., Chen, D., Lee, N., & Fung, P. (2024). Measuring Political Bias in Large Language Models: What Is Said and How It Is Said. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 11142–11159.
  2. See, e.g., recent work using large language models to extract stance, affect, and agreement from political discourse, paired with rule-based scoring systems for affective polarization. The methodological pattern — LLM-as-judge for affective signal, validated against human-coded subsets — is now standard in the computational social-science literature.
  3. Santurkar, S., Durmus, E., Ladhak, F., Lee, C., Liang, P., & Hashimoto, T. (2023). Whose Opinions Do Language Models Reflect? Stanford HAI / OpinionQA. Demonstrates the use of public-opinion benchmarks as peer-set references for evaluating model output distributions.
  4. Aaker, J. L. (1997). Dimensions of Brand Personality. Journal of Marketing Research, 34(3), 347–356. See also Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A Model of (Often Mixed) Stereotype Content: Competence and Warmth Respectively Follow from Perceived Status and Competition. Journal of Personality and Social Psychology, 82(6), 878–902. The warmth-and-competence framework is the most widely validated taxonomy of social and brand affective perception, and generalizes naturally to political figures and advocacy organizations.
  5. Röttger, P., et al. IssueBench (MilaNLP, Bocconi University). Documents persistent issue-level bias patterns in state-of-the-art LLMs that are robust to standard prompting interventions, supporting the structural-residual claim made here.
About the Author
Ed Forman

Ed Forman is the founder of Raise Presence LLC, which built Kyanos to measure and improve how progressive candidates and causes are represented across AI platforms. About Raise Presence →