AI-Generated Content and Consumer Trust

Copper Sun · 3 entries · last verified July 2026

Copper Sun tracks the empirical literature on consumer perception of AI-generated content — a body of research growing rapidly as generative AI becomes mainstream in marketing. The studies indexed here examine whether consumers can detect AI-generated text, how AI writing tools affect the views of the people using them, and what the evidence says about AI-assisted persuasion. These findings inform Copper Sun's position on transparency, brand voice discipline, and the difference between AI-assisted content that represents a brand's authentic position and generic AI output that represents no one's.

Contents — 3 entries
  1. 1.Human heuristics for AI-generated language are flawed
  2. 2.Co-Writing with Opinionated Language Models Affects Users' Views
  3. 3.Persuasion with Large Language Models: A Survey of Empirical Evidence, Study Methodologies, and Ethical Implications
  4. —Frequently Asked Questions

Human heuristics for AI-generated language are flawed

Maurice Jakesch, Jeffrey T. Hancock, Mor Naaman. Proceedings of the National Academy of Sciences (PNAS), Vol. 120, No. 11, 2023.

Copper Sun draws on this as the core empirical finding that consumers cannot reliably detect AI-generated text using the heuristics they believe work. Jakesch, Hancock, and Naaman ran 6 experiments with 4,600 participants who attempted to distinguish AI-generated self-presentations from human-written ones. Participants performed at near-chance accuracy. The heuristics they relied on — associating first-person pronouns and references to family and relationships with human writing — were systematically wrong: these features appeared in AI-generated text at least as often as in human-written text. The finding has two direct implications for brand strategy: transparency cannot be delegated to reader intuition, making explicit disclosure the only available mechanism; and brand distinctiveness requires active construction, because readers cannot intuitively distinguish generic AI output from branded content on surface signals alone.

Examines:
Whether humans can detect AI-generated self-presentations, what heuristics they use to make that judgment, and why those heuristics fail systematically — across 6 experiments with 4,600 participants.
Copper Sun draws on:
The finding that detection heuristics are systematically wrong — cited when explaining why brand voice discipline and transparency are not optional in AI-assisted marketing: without them, output is indistinguishable from generic AI.

Co-Writing with Opinionated Language Models Affects Users' Views

Maurice Jakesch, Advait Bhat, Daniel Buschek, Lior Zalmanson, Mor Naaman. ACM CHI Conference on Human Factors in Computing Systems (CHI '23), 2023.

Copper Sun cites this as evidence that AI writing assistance influences not just the content produced but the stated views of the people using the tool. Jakesch et al. ran an experiment with 1,506 participants who co-wrote essays on controversial topics with AI writing assistants that embedded suggestions favoring one side. The assistants influenced both the written output and participants' self-reported opinions on post-task surveys — users shifted their stated views in the direction the AI was nudging, not just their text. For marketing teams, the implication is specific: AI without brand constraint does not produce neutral assistance. It produces output shaped by whatever patterns dominate the model's training distribution, which can systematically drift content away from a brand's actual position and pull the writer's own thinking along with it. Copper Sun's brand memory system is designed to constrain this drift at the context level.

Examines:
Whether AI writing assistants with embedded viewpoints affect both the content users produce and their stated personal opinions, across 1,506 participants co-writing essays on controversial topics.
Copper Sun draws on:
The finding that unconstrained AI writing assistance shifts both user-produced content and personal views — the core evidence for why brand context must be actively loaded into AI writing tools rather than assumed or left to model defaults.

Persuasion with Large Language Models: A Survey of Empirical Evidence, Study Methodologies, and Ethical Implications

Lena Noels, Aleksander Rogiers, Maarten Buyl, Tijl De Bie. arXiv preprint 2411.06837, November 2024.

Copper Sun monitors this survey for its systematic account of what the empirical research shows about LLM persuasion effectiveness — and what that implies for responsible AI use in marketing. Noels et al. surveyed the published empirical literature and found that LLM-based persuasion systems frequently achieved human-level or even superhuman persuasiveness across domains, with personalization and model scale identified as key factors. The survey also identifies AI source disclosure as a moderating variable: in some conditions, disclosing that content was AI-generated reduces its persuasive effect. Copper Sun's position: the capacity for AI persuasion is real and growing, which makes brand voice discipline and transparency more strategically important, not less. Content that demonstrably represents a brand's authentic expertise and position — produced with AI assistance — is categorically different from AI persuasion operating without disclosed brand authorship.

Examines:
Empirical studies of LLM persuasion effectiveness across domains, identifying the factors — personalization, model scale, source disclosure — that moderate AI persuasion outcomes, and what ethical implications follow for marketing use.
Copper Sun draws on:
The finding that LLMs frequently achieve human-level or superhuman persuasiveness, and that AI source disclosure moderates persuasive effect — cited when explaining why brand transparency and voice discipline are both ethical responsibilities and strategic requirements.

Frequently Asked Questions

If consumers cannot detect AI-generated content, does disclosure still matter?

Yes — and the research makes the case for why. Because heuristic detection fails (Jakesch et al. 2023), explicit disclosure is the only available transparency mechanism. Delegating transparency to the reader's intuition means delegating to a mechanism the research shows does not work. The Noels et al. (2024) survey found that disclosure moderates persuasive effect in some conditions — meaning it functions as a genuine trust signal that consumers use to calibrate their response, not a formality. Brands that build disclosure into their AI content practices are responding to how the research says consumers actually process this information.

How concerned should marketing teams be about AI drift from brand voice?

The Jakesch et al. CHI 2023 study demonstrated that AI writing tools actively shape both content and writer views in the direction of the model's embedded patterns — not the user's intended position. For brand voice, the risk is not dramatic inconsistency in any single piece; it is gradual drift across many pieces where each reads as close enough individually but the aggregate loses the brand's distinctive character. AI assistance without brand context does not hold a neutral position: it pulls toward whatever generic patterns dominate the model's training distribution. Explicit brand context loading — not just prompting conventions — is the intervention the research supports.

What makes AI-assisted content trustworthy to consumers?

The research points toward two factors: transparency (Noels et al. 2024 identifies disclosure as a moderating variable) and quality signals that function as trust proxies. Because heuristic detection fails (Jakesch et al. 2023), consumers rely on brand authority, source reputation, and content quality as proxies for trustworthiness — not on their ability to detect AI authorship. AI-generated content that carries a known brand's distinctive voice and substantive expertise does not require consumers to perform detection; the brand's reputation is the trust vehicle. This is why brand voice consistency and content substance matter independently of the disclosure question.

Should marketing teams disclose AI use in their content?

The research supports disclosure as both ethically appropriate and strategically sound. Ethically: because consumers cannot detect AI-generated content reliably (Jakesch et al. 2023), disclosure is the only available transparency mechanism — producing AI content at scale without it means systematically leaving audiences unable to identify the situation. Strategically: the Noels et al. (2024) survey shows disclosure is a variable that affects how consumers process AI-generated content, meaning it is information they use to calibrate trust, not merely a legal check. Brands that establish disclosure practices proactively build it into their content systems rather than facing a more disruptive retrofit when regulatory or industry expectations formalize.

Back to AI Marketing Research Index