{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T15:10:22Z","timestamp":1788966622872,"version":"build-2803163510"},"reference-count":76,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/tdm\/userlicense\/1.0\/"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/www.elsevier.com\/legal\/tdmrep-license"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-017"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-012"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2027,2,1]],"date-time":"2027-02-01T00:00:00Z","timestamp":1801440000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-004"}],"content-domain":{"domain":["elsevier.com","sciencedirect.com"],"crossmark-restriction":true},"short-container-title":["Current Opinion in Psychology"],"published-print":{"date-parts":[[2027,2]]},"DOI":"10.1016\/j.copsyc.2026.102412","type":"journal-article","created":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T10:50:05Z","timestamp":1788951005000},"page":"102412","update-policy":"https:\/\/doi.org\/10.1016\/elsevier_cm_policy","source":"Crossref","is-referenced-by-count":0,"special_numbering":"C","title":["Artificial intelligence systems distort upstream selection in human social learning"],"prefix":"10.1016","volume":"73","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-6075-5446","authenticated-orcid":false,"given":"William J.","family":"Brady","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Chen-Wei","family":"Yu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Nicholas","family":"Ornstein","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Bolun","family":"Sun","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Joshua Conrad","family":"Jackson","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"key":"10.1016\/j.copsyc.2026.102412_bib1","series-title":"The secret of our success: how culture is driving human evolution, domesticating our species, and making us smarter","author":"Henrich","year":"2016"},{"key":"10.1016\/j.copsyc.2026.102412_bib2","series-title":"Not by genes alone: how culture transformed human evolution","author":"Richerson","year":"2010"},{"key":"10.1016\/j.copsyc.2026.102412_bib3","doi-asserted-by":"crossref","first-page":"123","DOI":"10.1002\/evan.10110","article-title":"The evolution of cultural evolution","volume":"12","author":"Henrich","year":"2003","journal-title":"Evol Anthropol Issues News Rev"},{"key":"10.1016\/j.copsyc.2026.102412_bib4","series-title":"Cultural evolution: how Darwinian theory can explain human culture and synthesize the social sciences","author":"Mesoudi","year":"2011"},{"key":"10.1016\/j.copsyc.2026.102412_bib5","doi-asserted-by":"crossref","first-page":"651","DOI":"10.1016\/j.tics.2018.04.003","article-title":"Social learning strategies: Bridge-building between fields","volume":"22","author":"Kendal","year":"2018","journal-title":"Trends Cognit Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib6","doi-asserted-by":"crossref","first-page":"215","DOI":"10.1016\/S1090-5138(98)00018-X","article-title":"The evolution of conformist transmission and the emergence of between-group differences","volume":"19","author":"Henrich","year":"1998","journal-title":"Evol Hum Behav"},{"key":"10.1016\/j.copsyc.2026.102412_bib7","doi-asserted-by":"crossref","first-page":"165","DOI":"10.1016\/S1090-5138(00)00071-4","article-title":"The evolution of prestige: freely conferred deference as a mechanism for enhancing the benefits of cultural transmission","volume":"22","author":"Henrich","year":"2001","journal-title":"Evol Hum Behav"},{"key":"10.1016\/j.copsyc.2026.102412_bib8","doi-asserted-by":"crossref","first-page":"1855","DOI":"10.1038\/s41562-023-01742-2","article-title":"Machine culture","volume":"7","author":"Brinkmann","year":"2023","journal-title":"Nat Hum Behav"},{"key":"10.1016\/j.copsyc.2026.102412_bib9","doi-asserted-by":"crossref","first-page":"947","DOI":"10.1016\/j.tics.2023.06.008","article-title":"Algorithm-mediated social learning in online social networks","volume":"27","author":"Brady","year":"2023","journal-title":"Trends Cognit Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib10","first-page":"5","article-title":"Large AI models have a prioritization problem: policy implications and solutions, policy insights","volume":"13","author":"Jackson","year":"2026","journal-title":"Behav Brain Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib11","doi-asserted-by":"crossref","DOI":"10.1016\/j.tics.2026.01.003","article-title":"The homogenizing effect of large language models on human expression and thought","author":"Sourati","year":"2026","journal-title":"Trends Cognit Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib12","doi-asserted-by":"crossref","first-page":"181","DOI":"10.1080\/00913367.2013.827606","article-title":"Following celebrities\u2019 tweets about brands: the impact of twitter-based electronic word-of-mouth on consumers\u2019 source credibility perception, buying intention, and social identification with celebrities","volume":"43","author":"Jin","year":"2014","journal-title":"J Advert"},{"key":"10.1016\/j.copsyc.2026.102412_bib13","doi-asserted-by":"crossref","first-page":"90","DOI":"10.1016\/j.obhdp.2018.12.005","article-title":"Algorithm appreciation: people prefer algorithmic to human judgment","volume":"151","author":"Logg","year":"2019","journal-title":"Organ Behav Hum Decis Process"},{"key":"10.1016\/j.copsyc.2026.102412_bib14","doi-asserted-by":"crossref","first-page":"49","DOI":"10.1038\/s41586-024-07146-0","article-title":"Artificial intelligence and illusions of understanding in scientific research","volume":"627","author":"Messeri","year":"2024","journal-title":"Nature"},{"key":"10.1016\/j.copsyc.2026.102412_bib15","series-title":"Adv. Exp. Soc. Psychol.","doi-asserted-by":"crossref","first-page":"161","DOI":"10.1016\/S0065-2601(08)60238-5","article-title":"Pluralistic ignorance and the perpetuation of social norms by unwitting actors","author":"Prentice","year":"1996"},{"key":"10.1016\/j.copsyc.2026.102412_bib16","doi-asserted-by":"crossref","DOI":"10.1525\/collabra.124533","article-title":"An illusory consensus effect: the mere repetition of information increases estimates that others would believe or already know it","volume":"10","author":"Jalbert","year":"2024","journal-title":"Collabra Psychol"},{"key":"10.1016\/j.copsyc.2026.102412_bib17","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0276367","article-title":"Longitudinal analysis of sentiment and emotion in news media headlines using automated labelling with transformer language models","volume":"17","author":"Rozado","year":"2022","journal-title":"PLoS One"},{"key":"10.1016\/j.copsyc.2026.102412_bib18","doi-asserted-by":"crossref","first-page":"1643","DOI":"10.1038\/s41562-024-01959-9","article-title":"How large language models can reshape collective intelligence","volume":"8","author":"Burton","year":"2024","journal-title":"Nat Hum Behav"},{"key":"10.1016\/j.copsyc.2026.102412_bib19","doi-asserted-by":"crossref","first-page":"746","DOI":"10.1037\/xge0000673","article-title":"Attentional capture helps explain why moral and emotional content go viral","volume":"149","author":"Brady","year":"2020","journal-title":"J Exp Psychol Gen"},{"key":"10.1016\/j.copsyc.2026.102412_bib20","doi-asserted-by":"crossref","first-page":"22","DOI":"10.1016\/j.cognition.2014.02.007","article-title":"The moral pop-out effect: enhanced perceptual awareness of morally relevant stimuli","volume":"132","author":"Gantman","year":"2014","journal-title":"Cognition"},{"key":"10.1016\/j.copsyc.2026.102412_bib21","doi-asserted-by":"crossref","DOI":"10.1073\/pnas.2024292118","article-title":"Out-group animosity drives engagement on social media","volume":"118","author":"Rathje","year":"2021","journal-title":"Proc Natl Acad Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib22","first-page":"1","article-title":"Negativity drives online news consumption","author":"Robertson","year":"2023","journal-title":"Nat Hum Behav"},{"key":"10.1016\/j.copsyc.2026.102412_bib23","series-title":"Stand out of our light: freedom and resistance in the attention economy","author":"Williams","year":"2018"},{"key":"10.1016\/j.copsyc.2026.102412_bib24","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1017\/S0003055421000885","article-title":"The psychology of online political hostility: a comprehensive, cross-national test of the mismatch hypothesis","volume":"116","author":"Bor","year":"2022","journal-title":"Am Polit Sci Rev"},{"key":"10.1016\/j.copsyc.2026.102412_bib25","doi-asserted-by":"crossref","first-page":"976","DOI":"10.1177\/09567976241258149","article-title":"Why Twitter sometimes rewards what most people disapprove of: the case of cross-party political relations","volume":"35","author":"Heltzel","year":"2024","journal-title":"Psychol Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib26","series-title":"The tip of the iceberg: how the social media production-consumption gap distorts public opinion for citizens and researchers","author":"Oswald","year":"2026"},{"key":"10.1016\/j.copsyc.2026.102412_bib27","article-title":"Inside the funhouse mirror factory: how social media distorts perceptions of norms","volume":"60","author":"Robertson","year":"2024","journal-title":"Curr Opin Psychol"},{"key":"10.1016\/j.copsyc.2026.102412_bib28","doi-asserted-by":"crossref","first-page":"3606","DOI":"10.1016\/j.vaccine.2018.05.040","article-title":"Polarization of the vaccination debate on Facebook","volume":"36","author":"Schmidt","year":"2018","journal-title":"Vaccine"},{"key":"10.1016\/j.copsyc.2026.102412_bib29","doi-asserted-by":"crossref","first-page":"230","DOI":"10.1038\/s41586-020-2281-1","article-title":"The online competition between pro- and anti-vaccination views","volume":"582","author":"Johnson","year":"2020","journal-title":"Nature"},{"key":"10.1016\/j.copsyc.2026.102412_bib30","series-title":"Culture and the evolutionary process","author":"Boyd","year":"1988"},{"key":"10.1016\/j.copsyc.2026.102412_bib31","doi-asserted-by":"crossref","DOI":"10.1371\/journal.pone.0255346","article-title":"Trusting the experts: the domain-specificity of prestige-biased social learning","volume":"16","author":"Brand","year":"2021","journal-title":"PLoS One"},{"key":"10.1016\/j.copsyc.2026.102412_bib32","doi-asserted-by":"crossref","first-page":"991","DOI":"10.1126\/science.adl2829","article-title":"Misinformation exploits outrage to spread online","volume":"386","author":"McLoughlin","year":"2024","journal-title":"Science"},{"key":"10.1016\/j.copsyc.2026.102412_bib33","doi-asserted-by":"crossref","first-page":"354","DOI":"10.1126\/science.adz1697","article-title":"How malicious AI swarms can threaten democracy","volume":"391","author":"Schroeder","year":"2026","journal-title":"Science"},{"key":"10.1016\/j.copsyc.2026.102412_bib34","series-title":"Towards measuring the representation of subjective global opinions in language models","author":"Durmus","year":"2024"},{"key":"10.1016\/j.copsyc.2026.102412_bib35","series-title":"Int. Conf. Mach. Learn.","first-page":"29971","article-title":"Whose opinions do language models reflect?","author":"Santurkar","year":"2023"},{"key":"10.1016\/j.copsyc.2026.102412_bib36","doi-asserted-by":"crossref","DOI":"10.1038\/s41598-025-98955-4","article-title":"Prestige bias drives the viral spread of content reposted by influencers in online communities","volume":"15","author":"Niitsuma","year":"2025","journal-title":"Sci Rep"},{"key":"10.1016\/j.copsyc.2026.102412_bib37","doi-asserted-by":"crossref","first-page":"189","DOI":"10.1007\/s40806-020-00261-x","article-title":"The cultural transmission of prestige and dominance social rank cues: an experimental simulation","volume":"7","author":"Jim\u00e9nez","year":"2021","journal-title":"Evol Psychol Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib38","series-title":"Creat. Cogn.","first-page":"413","article-title":"Homogenization effects of large language models on human creative ideation","author":"Anderson","year":"2024"},{"key":"10.1016\/j.copsyc.2026.102412_bib39","doi-asserted-by":"crossref","first-page":"2154","DOI":"10.1080\/01621459.2025.2555067","article-title":"On the algorithmic bias of aligning large language models with RLHF: preference collapse and matching regularization","volume":"120","author":"Xiao","year":"2025","journal-title":"J Am Stat Assoc"},{"key":"10.1016\/j.copsyc.2026.102412_bib40","series-title":"Understanding the effects of RLHF on LLM generalisation and diversity","author":"Kirk","year":"2024"},{"key":"10.1016\/j.copsyc.2026.102412_bib41","series-title":"Int. Conf. Learn. Represent.","first-page":"642","article-title":"Does writing with language models reduce content diversity?","author":"Padmakumar","year":"2024"},{"key":"10.1016\/j.copsyc.2026.102412_bib42","series-title":"A rational analysis of the effects of sycophantic AI","author":"Batista","year":"2026"},{"key":"10.1016\/j.copsyc.2026.102412_bib43","doi-asserted-by":"crossref","DOI":"10.1126\/science.aec8352","article-title":"Sycophantic AI decreases prosocial intentions and promotes dependence","volume":"391","author":"Cheng","year":"2026","journal-title":"Science"},{"key":"10.1016\/j.copsyc.2026.102412_bib44","series-title":"Sycophantic AI increases attitude extremity and overconfidence","author":"Rathje","year":"2026"},{"key":"10.1016\/j.copsyc.2026.102412_bib45","doi-asserted-by":"crossref","first-page":"9020","DOI":"10.1073\/pnas.1008636108","article-title":"How social influence can undermine the wisdom of crowd effect","volume":"108","author":"Lorenz","year":"2011","journal-title":"Proc Natl Acad Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib46","series-title":"Proc. 2021 ACM conf. Fairness account. Transpar., ACM","first-page":"610","article-title":"On the dangers of stochastic parrots: can language models be too big?","author":"Bender","year":"2021"},{"key":"10.1016\/j.copsyc.2026.102412_bib47","first-page":"8424","article-title":"Deduplicating training data makes language models better","volume":"vol. 1","author":"Lee","year":"2022"},{"key":"10.1016\/j.copsyc.2026.102412_bib48","series-title":"Quantifying memorization across neural language models","author":"Carlini","year":"2023"},{"key":"10.1016\/j.copsyc.2026.102412_bib49","doi-asserted-by":"crossref","first-page":"16385","DOI":"10.1073\/pnas.0403723101","article-title":"Groups of diverse problem solvers can outperform groups of high-ability problem solvers","volume":"101","author":"Hong","year":"2004","journal-title":"Proc Natl Acad Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib50","article-title":"Homogenizing effect of large language models (LLMs) on creative diversity: an empirical comparison of human and ChatGPT writing","volume":"6","author":"Moon","year":"2025","journal-title":"Comput Hum Behav Artif Hum"},{"key":"10.1016\/j.copsyc.2026.102412_bib51","doi-asserted-by":"crossref","DOI":"10.1093\/pnasnexus\/pgad325","article-title":"The social media discourse of engaged partisans is toxic even when politics are irrelevant","volume":"2","author":"Mamakos","year":"2023","journal-title":"PNAS Nexus"},{"key":"10.1016\/j.copsyc.2026.102412_bib52","doi-asserted-by":"crossref","DOI":"10.1093\/pnasnexus\/pgaf062","article-title":"Engagement, user satisfaction, and the amplification of divisive content on social media","volume":"4","author":"Milli","year":"2025","journal-title":"PNAS Nexus"},{"key":"10.1016\/j.copsyc.2026.102412_bib53","first-page":"1","article-title":"Redesigning algorithms to intervene on social norm misperceptions during a national election","author":"Brady","year":"2026","journal-title":"Nature"},{"key":"10.1016\/j.copsyc.2026.102412_bib54","doi-asserted-by":"crossref","first-page":"416","DOI":"10.1038\/s41586-026-10098-2","article-title":"The political effects of x\u2019s feed algorithm","volume":"652","author":"Gauthier","year":"2026","journal-title":"Nature"},{"key":"10.1016\/j.copsyc.2026.102412_bib55","series-title":"Engagement-based algorithms disrupt human social norm learning","author":"Brady","year":"2025"},{"key":"10.1016\/j.copsyc.2026.102412_bib56","doi-asserted-by":"crossref","first-page":"831","DOI":"10.1257\/aer.20191777","article-title":"Social media, news consumption, and polarization: evidence from a field experiment","volume":"111","author":"Levy","year":"2021","journal-title":"Am Econ Rev"},{"key":"10.1016\/j.copsyc.2026.102412_bib57","series-title":"Break. Soc. Media prism","article-title":"Breaking the social media prism: how to make our platforms less polarizing","author":"Bail","year":"2022"},{"key":"10.1016\/j.copsyc.2026.102412_bib58","doi-asserted-by":"crossref","first-page":"922","DOI":"10.1093\/joc\/jqab034","article-title":"The distorting prism of social media: how self-selection and exposure to incivility fuel online comment toxicity","volume":"71","author":"Kim","year":"2021","journal-title":"J Commun"},{"key":"10.1016\/j.copsyc.2026.102412_bib59","doi-asserted-by":"crossref","first-page":"4779","DOI":"10.1038\/s41467-022-32412-y","article-title":"Americans experience a false social reality by underestimating popular climate policy support by nearly half","volume":"13","author":"Sparkman","year":"2022","journal-title":"Nat Commun"},{"key":"10.1016\/j.copsyc.2026.102412_bib60","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1111\/j.1460-2466.1974.tb00367.x","article-title":"The spiral of silence: a theory of public opinion","volume":"24","author":"Noelle-Neumann","year":"1974","journal-title":"J Commun"},{"key":"10.1016\/j.copsyc.2026.102412_bib61","series-title":"Rising moralization in social media discourse","author":"Puryear","year":"2026"},{"key":"10.1016\/j.copsyc.2026.102412_bib62","series-title":"Proc. 2023 CHI conf. Hum. Factors comput. Syst.","first-page":"1","article-title":"Co-writing with opinionated language models affects users\u2019 views","author":"Jakesch","year":"2023"},{"key":"10.1016\/j.copsyc.2026.102412_bib63","series-title":"The basic b\u2217\u2217\u2217 effect: the use of LLM-based agents reduces the distinctiveness and diversity of people\u2019s choices","author":"Matz","year":"2025"},{"key":"10.1016\/j.copsyc.2026.102412_bib64","doi-asserted-by":"crossref","DOI":"10.1073\/pnas.2116851119","article-title":"Correcting inaccurate metaperceptions reduces Americans\u2019 support for partisan violence","volume":"119","author":"Mernyk","year":"2022","journal-title":"Proc Natl Acad Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib65","doi-asserted-by":"crossref","first-page":"742","DOI":"10.1037\/pspi0000516","article-title":"Overestimating the social costs of political belief change","volume":"130","author":"Spelman","year":"2026","journal-title":"J Pers Soc Psychol"},{"key":"10.1016\/j.copsyc.2026.102412_bib66","series-title":"Biased motive inferences of online outrage undermine cross-partisan conversations","author":"Yu","year":"2025"},{"key":"10.1016\/j.copsyc.2026.102412_bib67","doi-asserted-by":"crossref","DOI":"10.1073\/pnas.1912440117","article-title":"Why the backfire effect does not explain the durability of political misperceptions","volume":"118","author":"Nyhan","year":"2021","journal-title":"Proc Natl Acad Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib68","series-title":"Verbalized sampling: how to mitigate mode collapse and unlock LLM diversity","author":"Zhang","year":"2025"},{"key":"10.1016\/j.copsyc.2026.102412_bib69","doi-asserted-by":"crossref","DOI":"10.1002\/bdm.70055","article-title":"Similarity and consistency in algorithm-guided exploration","volume":"38","author":"Danwitz","year":"2025","journal-title":"J Behav Decis Making"},{"key":"10.1016\/j.copsyc.2026.102412_bib70","doi-asserted-by":"crossref","DOI":"10.1016\/j.tics.2026.06.009","article-title":"Misinformation as strategy: epistemic consequences and the undermining of shared truth","author":"Ecker","year":"2026","journal-title":"Trends Cognit Sci"},{"key":"10.1016\/j.copsyc.2026.102412_bib71","unstructured":"Perspective arguing that AI tools can create \u201cillusions of understanding\u201d \u2014 researchers feel they grasp more than they do \u2014 and warning that broad AI adoption risks producing epistemic monocultures that narrow the questions and methods a field explores. Importance: it grounds the that users grant model outputs unearned, prestige-like deference, treating a single confident answer as if it were vouched for by collective expertise."},{"key":"10.1016\/j.copsyc.2026.102412_bib72","unstructured":"Across multiple studies, misinformation sources evoked more moral outrage than trustworthy sources, and outrage in turn drove sharing \u2014 including a willingness to share outrage-inducing falsehoods that people hadn\u2019t read carefully or fully believed. Importance: provides a viable mechanism as to why engagement optimization preferentially elevates a hostile, misleading minority and how skill in capturing attention can get mistaken for competence."},{"key":"10.1016\/j.copsyc.2026.102412_bib73","unstructured":"An empirical comparison of human writing versus ChatGPT-assisted writing showing that ChatGPT users generated less semantically diverse ideas, with homogenization appearing at the group level even though individuals subjectively felt original. Importance: direct evidence that mode collapse propagates from model to people. The diversity loss is invisible for users, which is exactly what makes it dangerous for social learning."},{"key":"10.1016\/j.copsyc.2026.102412_bib74","unstructured":"A preregistered audit of Twitter\u2019s ranking found that, versus a chronological baseline, the engagement-based algorithm amplified emotionally charged, out-group\u2013hostile content. It also found that users didn\u2019t actually prefer the tweets the engagement algorithm selected. Importance: shows tail-amplification is driven by the optimization target, not genuine user demand, strengthening the claim that the engagement signal itself corrupts the upstream sample and that re-ranking could be a viable solution."},{"key":"10.1016\/j.copsyc.2026.102412_bib75","unstructured":"A preregistered field experiment that assigned 2000 participants to custom feed-ranking algorithms for eight weeks around the 2024 U.S. election. The engagement-based algorithm amplified moralized, emotional, and toxic content, degraded the accuracy of users\u2019 perceptions of social norms, and inflated perceived partisan animosity; an intervention algorithm that down-weighted extreme users reduced divisive content and produced the most accurate norm perceptions. Importance: clear causal demonstration that upstream ranking directly distorts what people see and infer about the majority, and it shows the distortion could potentially be improved by alternative algorithms."},{"key":"10.1016\/j.copsyc.2026.102412_bib76","unstructured":"Introduces a prompting strategy that asks a model for several distinct responses rather than one modal output, partially recovering the output diversity that standard alignment training suppresses. Example of how the \u201cdiversification of AI outputs\u201d intervention may be concrete and tractable, showing mode collapse is at least partly a surface-level training\/decoding artifact that can be counteracted without retraining."}],"container-title":["Current Opinion in Psychology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S2352250X26001508?httpAccept=text\/xml","content-type":"text\/xml","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/api.elsevier.com\/content\/article\/PII:S2352250X26001508?httpAccept=text\/plain","content-type":"text\/plain","content-version":"vor","intended-application":"text-mining"}],"deposited":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T14:16:29Z","timestamp":1788963389000},"score":1,"resource":{"primary":{"URL":"https:\/\/linkinghub.elsevier.com\/retrieve\/pii\/S2352250X26001508"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2027,2]]},"references-count":76,"alternative-id":["S2352250X26001508"],"URL":"https:\/\/doi.org\/10.1016\/j.copsyc.2026.102412","relation":{},"ISSN":["2352-250X"],"issn-type":[{"value":"2352-250X","type":"print"}],"subject":[],"published":{"date-parts":[[2027,2]]},"assertion":[{"value":"Elsevier","name":"publisher","label":"This article is maintained by"},{"value":"Artificial intelligence systems distort upstream selection in human social learning","name":"articletitle","label":"Article Title"},{"value":"Current Opinion in Psychology","name":"journaltitle","label":"Journal Title"},{"value":"https:\/\/doi.org\/10.1016\/j.copsyc.2026.102412","name":"articlelink","label":"CrossRef DOI link to publisher maintained version"},{"value":"article","name":"content_type","label":"Content Type"},{"value":"\u00a9 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.","name":"copyright","label":"Copyright"}],"article-number":"102412"}}