Many people now use AI chatbots to obtain summaries of complex topics, yet we know little about how this affects knowledge acquisition, including how the effects might vary across different groups of people. We conducted two experiments comparing how well people recalled factual information after reading AI-generated or human-written historical summaries. Participants who read AI-generated summaries scored significantly higher on knowledge tests than those who read expert-written blog posts (Study 1) or Wikipedia articles (Study 2). These improvements were present regardless of whether readers knew the content was AI-generated or if the AI summaries were politically biased. Moreover, AI summaries improved recall across various demographic groups, including gender, race, income, education, and digital literacy levels. This suggets that using AI tools for everyday factual queries does not create new knowledge inequalities but could still amplify existing ones through differential access. Our findings indicate that the increasingly routine use of AI for information-seeking could enhance factual learning, with implications for education policy and addressing inequality.
AbidA.FarooqiM.ZouJ. (2021). Persistent anti-Muslim bias in large language models. In Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society, Virtual Event, Association for Computational Machinery, pp. 298–306.
2.
AltayS.GilardiF. (2024). People are skeptical of headlines labeled as AI-generated, even if true or human-made, because they assume full AI automation. PNAS Nexus, 3(10), pgae403. https://doi.org/10.1093/pnasnexus/pgae403
3.
ArgyleL. P.BailC. A.BusbyE. C.GublerJ. R.HoweT.RyttingC.SorensenT.WingateD. (2023). Leveraging AI for democratic discourse: Chat interventions can improve online political conversations at scale. Proceedings of the National Academy of Sciences of the United States of America, 120(41), Article e2311627120. https://doi.org/10.1073/pnas.2311627120
4.
ArgyleL. P.BusbyE. C.FuldaN.RyttingC.WingateD. (2023). Out of one, many: Using language models to simulate human samples. Political Analysis, 31(3), 337–351. https://doi.org/10.1017/pan.2023.2
5.
AslettK.SandersonZ.GodelW.PersilyN.NaglerJ.TuckerJ. A. (2023). Online searches to evaluate misinformation can increase its perceived veracity. Nature, 625(7995), 1–9. https://doi.org/10.1038/s41586-023-06883-y
6.
BaiH.VoelkelJ. G.MuldowneyS.EichstaedtJ. C.WillerR. (2025). LLM-generated messages can persuade humans on policy issues. Nature Communications, 16(1), 6037. https://doi.org/10.1038/s41467-025-61345-5
7.
BaiX.WangA.SucholutskyI.GriffithsT. L. (2025). Explicitly unbiased large language models still form biased associations. Proceedings of the National Academy of Sciences of the United States of America, 122(8), Article e2416228122. https://doi.org/10.1073/pnas.2416228122
8.
BaldassarriD.GelmanA. (2008). Partisans without constraint: Political polarization and trends in American public opinion. AJS; American Journal of Sociology, 114(2), 408–446. https://doi.org/10.2139/ssrn.1010098
9.
BastaniH.BastaniO.SunguA.GeH.KabakcıÖ.MarimanR. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. Proceedings of the National Academy of Sciences of the United States of America, 122(26), Article e2422633122. https://doi.org/10.1073/pnas.2422633122
10.
BenderE. M.McMillan-MajorA.GebruT.SchmitchellS. (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, Virtual Event, Association for Computational Machinery, 610–623.
11.
BenjaminR. (2019). Race after technology: Abolitionist tools for the new Jim code. Polity.
12.
BianchiF.KalluriP.DurmusE.LadhakF.ChengM.NozzaD.HashimotoT.JurafskyD.ZouJ.CaliskanA. (2023). Easily accessible text-to-image generation amplifies demographic stereotypes at large scale. In Proceedings of the 2023 ACM conference on fairness, accountability, and transparency, Virtual Event, Association for Computational Machinery, 1493–1504.
13.
CapraroV.LentschA.AcemogluD.AkgunS.AkhmedovaA.BilanciniE.BonnefonJ. F.Brañas-GarzaP.ButeraL.DouglasK. M.EverettJ. A. C.GigerenzerG.GreenhowC.HashimotoD. A.Holt-LunstadJ.JettenJ.JohnsonS.KunzW. H.LongoniC.VialeR. (2024). The impact of generative artificial intelligence on socioeconomic inequalities and policy making. PNAS Nexus, 3(6), pgae191. https://doi.org/10.1093/pnasnexus/pgae191
14.
ClarkE.AugustT.SerranoS.HaduongN.GururanganS.SmithN. A. (2021). All that's 'human' is not gold: Evaluating human evaluation of generated text. In Proceedings of the 59th annual meeting of the association for computational linguistics and the 11th international joint conference on natural language processing, Association for Computational Linguistics, Online,7282–7296.
15.
CostelloT. H.PennycookG.RandD. G. (2024). Durably reducing conspiracy beliefs through dialogues with AI. Science, 385(6714), eadq1814. https://doi.org/10.1126/science.adq1814
16.
DathathriS.SeeA.GhaisasS.HuangP. S.McAdamR.WelblJ.BachaniV.KaskasoliA.StanforthR.MatejovicovaT.HayesJ.VyasN.Al MereyM.Brown-CohenJ.BunelR.BalleB.CemgilT.AhmedZ.StacpooleK.KohliP. (2024). Scalable watermarking for identifying large language model outputs. Nature, 634(8035), 818–823. https://doi.org/10.1038/s41586-024-08025-4
17.
DellaPostaD. (2020). Pluralistic collapse: The ‘oil spill’ model of mass opinion polarization. American Sociological Review, 85(3), 507–536. https://doi.org/10.1177/000312242092298
18.
DiMaggioP.EvansJ.BrysonB. (1996). Have American's social attitudes become more polarized?American Journal of Sociology, 102(3), 690–755. https://doi.org/10.1086/230995
19.
DoshiA. R.HauserO. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28), eadn5290. https://doi.org/10.1126/sciadv.adn5290
20.
FarquharS.KossenJ.KuhnL.GalY. (2024). Detecting hallucinations in large language models using semantic entropy. Nature, 630(8017), 625–630. https://doi.org/10.1038/s41586-024-07421-0
21.
FeuerriegelS.DiRestaR.GoldsteinJ. A.KumarS.Lorenz-SpreenP.TomzM.PröllochsN. (2023). Research can help to tackle AI-generated disinformation. Nature Human Behaviour, 7(11), 1818–1821. https://doi.org/10.1038/s41562-023-01726-2
22.
FinkelE. J.BailC. A.CikaraM.DittoP. H.IyengarS.KlarS.MasonL.McGrathM. C.NyhanB.RandD. G.SkitkaL. J.TuckerJ. A.Van BavelJ. J.WangC. S.DruckmanJ. N. (2020). Political sectarianism in America. Science, 370(6516), 533–536. https://doi.org/10.1126/science.abe1715
23.
GallegosI. O.ShaniC.ShiW.BianchiF.GainsburgI.JurafskyD.WillerR. (2025). Labeling messages as AI-Generated does not reduce their persuasive effects. arXiv. https://doi.org/10.48550/arXiv.2504.09865
24.
GelmanA.CarlinJ. (2014). Beyond power calculations: Assessing type S (sign) and type M (magnitude) errors. Perspectives on Psychological Science: A Journal of the Association for Psychological Science, 9(6), 641–651. https://doi.org/10.1177/1745691614551642
25.
GilardiF.Di LorenzoS.EzzainiJ.SantaB.StreiffB.ZurfluhE.HoesE. (2024). Disclosure of AI-generated news increases engagement but does not reduce aversion, despite positive quality ratings. arXiv. https://arxiv.org/pdf/2409.03500
GreensteinS.ZhuF. (2018). Do experts or crowd-based models produce more bias? Evidence from Encyclopedia Britannica and Wikipedia. MIS Quarterly, 42(3), 945–958. https://doi.org/10.25300/misq/2018/14084
28.
HackenburgK.MargettsH. (2024). Evaluating the persuasive influence of political microtargeting with large language models. Proceedings of the National Academy of Sciences, 121(24), Article e2403116121. https://doi.org/10.1073/pnas.2403116121
HällstenM.PfefferF. T. (2017). Grand advantage: Family wealth and grandchildren's educational achievement in Sweden. American Sociological Review, 82(2), 328–360. https://doi.org/10.1177/0003122417695791
31.
HargittaiE.MicheliM. (2019). Internet skills and why they matter. In GrahamMarkDuttonWilliam H. (Eds.), Society & the Internet: How networks of information and communication are changing our lives (pp. 109–124). Oxford University Press.
32.
HofmannV.KalluriP. R.JurafskyD.KingS. (2024). AI generates covertly racist decisions about people based on their dialect. Nature, 633(8028), 147–154. https://doi.org/10.1038/s41586-024-07856-5
33.
JoyceK.CruzT. M. (2024). A sociology of artificial intelligence: Inequalities, power, and data justice. Socius: Sociological Research for a Dynamic World, 10, Article 23780231241275393. https://doi.org/10.1177/23780231241275393
34.
JoyceK.Smith-DoerrL.AlegriaS.BellS.CruzT.HoffmanS. G.NobleS. U.ShestakofskyB. (2021). Toward a sociology of artificial intelligence: A call for research on inequalities and structural change. Socius: Sociological Research for a Dynamic World, 7, Article 2378023121999581. https://doi.org/10.1177/2378023121999581
35.
KarellD.SachsJ.BarrettR. (2025). Synthetic duality: A framework for analyzing generative artificial intelligence’s representation of social reality. Poetics, 108, 101966. https://doi.org/10.1016/j.poetic.2024.101966
36.
KrepsS.McCainR. M.BrundageM. (2022). All the news that's fit to fabricate: AI-generated text as a tool of media misinformation. Journal of Experimental Political Science, 9(1), 104–117. https://doi.org/10.1017/xps.2020.37
37.
LabaN. (2024). Engine for the imagination? Visual generative media and the issue of representation. Media, Culture & Society, 46(8), 1599–1620. https://doi.org/10.1177/01634437241259950
38.
LiaoW.WeismanW.ThakurA. (2024). On the motivations to seek information from artificial intelligence agents versus humans: A risk information seeking and processing perspective. Science Communication, 46(4), 458–486. https://doi.org/10.1177/10755470241232993
39.
MaN.ZhongZ. (2025). A meta-analysis of the impact of generative artificial intelligence on learning outcomes. Journal of Computer Assisted Learning, 41(5), Article e70117. https://doi.org/10.1111/jcal.70117
40.
MakhortykhM.VziatyshevaV.SydorovaM. (2023). Generative AI and contestation and instrumentalization of memory about the Holocaust in Ukraine. Eastern European Holocaust Studies, 1(2), 349–355. https://doi.org/10.1515/eehs-2023-0054
41.
McClainC.KennedyB.GottfriedJ.AndersonM.PasquiniG. (2025). How the U.S. public and AI experts view artificial intelligence. Pew Research Center.
42.
McCoskerA. (2024). Making sense of deepfakes: Socializing AI and building data literacy on GitHub and YouTube. New Media & Society, 26(5), 2786–2803. https://doi.org/10.1177/14614448221093943
43.
McKeeK. R.TacchettiA.BakkerM. A.BalaguerJ.Campbell-GillinghamL.EverettR.BotvinickM. (2023). Scaffolding cooperation in human groups with deep reinforcement learning. Nature Human Behaviour, 7(10), 1769–1787. https://doi.org/10.1038/s41562-023-01686-7
44.
MelumadS.YunJ. H. (2025). Experimental evidence of the effects of large language models versus web search on depth of learning. PNAS Nexus, 4(10), pgaf316. https://doi.org/10.1093/pnasnexus/pgaf316
MontgomeryJ. M.NyhanB.TorresM. (2018). How conditioning on posttreatment variables can ruin your experiment and what to do about it. American Journal of Political Science, 62(3), 760–775. https://doi.org/10.1111/ajps.12357
48.
NightingaleS. J.FaridH. (2022). AI-synthesized faces are indistinguishable from real faces and more trustworthy. Proceedings of the National Academy of Sciences of the United States of America, 119(8), Article e2120481119. https://doi.org/10.1073/pnas.2120481119
49.
NobleS. U. (2018). Algorithms of oppression: How search engines reinforce racism. NYU Press.
50.
OuyangL.WuJ.JiangX.AlmediaD.WainwrightC. L.MishkinP.ZhangC.AgarwalS.SlamaK.RayA.SchulmanJ.HiltonJ.KeltonF.MillerL.SimensM.AskellA.WelinderP.ChristianP.LeikeJ.LoweR. (2022). Training language models to follow instructions with human feedback. arXiv. https://arxiv.org/abs/2203.02155
51.
PadmakumarV.HeH. (2023). Does writing with language models reduce content diversity?arXiv, 2309.05196v1.
52.
PalmerA. K.SpirlingA. (2023). Large Language Models Can Argue in Convincing Ways About Politics, But Humans Dislike AI Authors: Implications for Governance. Political Science, 75(3), 281–291. DOI:10.1080/00323187.2024.2335471
53.
SpitaleG.Biller-AndornoN.GermaniF. (2023). AI model GPT-3 (dis)informs us better than humans. Science Advances, 9(26), eadh1850. https://doi.org/10.1126/sciadv.adh1850
54.
SteinertC. V.KazenwadelD. (2024). How user language affects conflict fatality estimates in ChatGPT. Journal of Peace Research, 62(4), 1128–1143. https://doi.org/10.1177/00223433241279381
55.
SteinssonS. (2024). Rule ambiguity, institutional clashes, and population loss: How Wikipedia became the last good place on the internet. American Political Science Review Review, 118(1), 235–251. https://doi.org/10.1017/s0003055423000138
UeshimaA.JonesM. I.ChristakisN. A. (2024). Simple autonomous agents can enhance creative semantic discovery by human groups. Nature Communications, 15(1), 5212. https://doi.org/10.1038/s41467-024-49528-y
59.
WasikB. (2025, June 16). A.I. is poised to rewrite history. Literally. The New York Times Magazine.
60.
YanL.GreiffS.TeuberZ.GrasevićD. (2024). Promises and challenges of generative artificial intelligence for human learning. Nature Human Behaviour, 8(10), 1839–1850. https://doi.org/10.1038/s41562-024-02004-5
61.
ZajkoM. (2022). Artificial intelligence, algorithms, and social inequality: Sociological contributions to contemporary debates. Sociology Compass, 16(3), Article e12962. https://doi.org/10.1111/soc4.12962
62.
ZhangY.GoslineR. (2023). Human favoritism, not AI aversion: People's perceptions (and bias) toward generative AI, human experts, and human–GAI collaboration in persuasive content generation. Judgment and Decision Making, 18, Article e41. https://doi.org/10.1017/jdm.2023.37
63.
ZhuY.LiuQ.ZhaoL. (2025). Exploring the impact of generative artificial intelligence on students’ learning outcomes: A meta-analysis. Education and Information Technologies Technologies, 30(11), 16211–16239. https://doi.org/10.1007/s10639-025-13420-z
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
