Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 79 of the 79 references it cites were found in a registry or an index.

Exploring the Use of Personalized AI for Identifying Misinformation on Social Media

Open access
Jahanbakhsh, Farnaz, Katsis, Yannis, Wang, Dakuo, Popa, Lucian, Muller, Michael · 2023 · Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems

Abstract

This work aims to explore how human assessments and AI predictions can be combined to identify misinformation on social media. To do so, we design a personalized AI which iteratively takes as training data a single user’s assessment of content and predicts how the same user would assess other content. We conduct a user study in which participants interact with a personalized AI that learns their assessments of a feed of tweets, shows its predictions of whether a user would find other tweets (in)accurate, and evolves according to the user feedback. We study how users perceive such an AI, and whether the AI predictions influence users’ judgment. We find that this influence does exist and it grows larger over time, but it is reduced when users provide reasoning for their assessment. We draw from our empirical observations to identify design implications and directions for future work.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1145/3544548.3581219

Cites

79 of 79 references verified0 citation errors · 0 not checkable

Statistics

Figures

Image 1
CC-BY · From the PDF
Figure 1 Flow of the steps of the user study.
Figure 1 Flow of the steps of the user study.CC-BY · From the PDF
Figure 2 Assess tweets without AI assistance (UI used in Steps 1 and 2). Participants assess each tweet in their feed, rate their level of confdence in their assessments, and provide reasoning for at least 3 of the assessments without seeing any AI predictions. In Step 2, the order of assessing tweets was predetermined, with the user being able to only assess the next few tweets marked with a blue arrow (not shown in the Figure).
Figure 2 Assess tweets without AI assistance (UI used in Steps 1 and 2). Participants assess each tweet in their feed, rate their level of confdence in their assessments, and provide reasoning for at least 3 of the assessments without seeing any AI predictions. In Step 2, the order of assessing tweets was predetermined, with the user being able to only assess the next few tweets marked with a blue arrow (not shown in the Figure).CC-BY · From the PDF
Figure 3 Assess tweets with AI assistance (UI used in Step 3). Participants assess each tweet in their feed while seeing AI model predictions. As shown in this screenshot, the user is in the middle of an experiment and has assessed the top tweet, while the rest of the tweets in this feed (for which the AI model predictions are shown) have not been assessed yet. The AI model predictions may change in response to user feedback. The tweets for which the AI predictions have recently changed are shown on the left pane. Newly changed predictions are diferentiated visually with a border and a notifcation icon, similar to the bottom tweet in the image.
Figure 3 Assess tweets with AI assistance (UI used in Step 3). Participants assess each tweet in their feed while seeing AI model predictions. As shown in this screenshot, the user is in the middle of an experiment and has assessed the top tweet, while the rest of the tweets in this feed (for which the AI model predictions are shown) have not been assessed yet. The AI model predictions may change in response to user feedback. The tweets for which the AI predictions have recently changed are shown on the left pane. Newly changed predictions are diferentiated visually with a border and a notifcation icon, similar to the bottom tweet in the image.CC-BY · From the PDF
Figure 4 The distribution of participants' responses to the question 'In general, how good was the AI at predicting your assessments?'.
Figure 4 The distribution of participants' responses to the question 'In general, how good was the AI at predicting your assessments?'.CC-BY · From the PDF
Figure 5 The distribution of participants' responses to the question 'How better did the AI get at predicting your assessments over time as you provided more assessments?'.
Figure 5 The distribution of participants' responses to the question 'How better did the AI get at predicting your assessments over time as you provided more assessments?'.CC-BY · From the PDF
Image 7
CC-BY · From the PDF
Image 8
CC-BY · From the PDF
Figure 8 The predicted values of user-model agreement by iteration. Model iteration is positively correlated with usermodel agreement, independent of which model. This shows that models become better at predicting users' assessments over time.
Figure 8 The predicted values of user-model agreement by iteration. Model iteration is positively correlated with usermodel agreement, independent of which model. This shows that models become better at predicting users' assessments over time.CC-BY · From the PDF
Figure 9 The effect of the interaction between model and iteration on the predicted user-model agreement. Over time, the difference is user-model agreement across the two models increases.
Figure 9 The effect of the interaction between model and iteration on the predicted user-model agreement. Over time, the difference is user-model agreement across the two models increases.CC-BY · From the PDF
Figure 10 Users ratings of how useful they fnd each of the content moderation strategies described in Section D on a 5 point likert scale. The means are rather close to each other and the standard deviations are very wide, indicating a high variance among users.
Figure 10 Users ratings of how useful they fnd each of the content moderation strategies described in Section D on a 5 point likert scale. The means are rather close to each other and the standard deviations are very wide, indicating a high variance among users.CC-BY · From the PDF
Figure 11 The participants' perceived usefulness of the content moderation strategies described in Section D.
Figure 11 The participants' perceived usefulness of the content moderation strategies described in Section D.CC-BY · From the PDF
Figure 12 The efect of the interaction between model and iteration on the predicted user-model agreement at iterations when the user was still in the Seeding step. Over time, the diference is user-model agreement across the two models increases. The diference in these initial iterations however, is smaller than the diference observed in Figure 9 that also includes the iterations after the Seeding step.
Figure 12 The efect of the interaction between model and iteration on the predicted user-model agreement at iterations when the user was still in the Seeding step. Over time, the diference is user-model agreement across the two models increases. The diference in these initial iterations however, is smaller than the diference observed in Figure 9 that also includes the iterations after the Seeding step.CC-BY · From the PDF
Image 14
CC-BY · From the PDF
Image 15
CC-BY · From the PDF
Figure 15 Instructions before condition Assisted
Figure 15 Instructions before condition AssistedCC-BY · From the PDF
Figure 14 Instructions before condition Unassisted.
Figure 14 Instructions before condition Unassisted.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

1 retracted

Show all 74 references

Research works printed but not linked: 14

  1. Mike Ananny. 2018. The partnership press: Lessons for platform-publisher collaborations as Facebook and news outlets team to fight misinformation. (2018).
  2. Adam J Berinsky, Gregory A Huber, and Gabriel S Lenz. 2012. Evaluating online labor markets for experimental research: Amazon. com’s Mechanical Turk. Political analysis 20, 3 (2012), 351–368.
  3. Monika Bickert. 2019. Combatting Vaccine Misinformation - About Facebook. Retrieved August 25, 2022 from https://about.fb.com/news/2019/03/combatting-vaccine-misinformation/
  4. Pranav Dixit and Ryan Mac. 2018. How WhatsApp Destroyed A Village. Buzzfeed News (2018).
  5. Yasmin Ibrahim. 2017. Facebook and the Napalm Girl: reframing the iconic as pornographic. Social Media+ Society 3, 4 (2017), 2056305117743140.
  6. Wolfgang Klein. 1994. Learning how to express temporality in a second language. In Società di linguistica Italiana, SLI 34: Italiano-lingua seconda/lingua straniera: Atti del XXVI Congresso. Bulzoni, 227–248.
  7. Pranav Malhotra. 2020. A Relationship-Centered and Culturally Informed Approach to Studying Misinformation on COVID-19. Social Media+ Society 6, 3 (2020), 2056305120948224.
  8. Drew B Margolin, Aniko Hannak, and Ingmar Weber. 2018. Political fact-checking on Twitter: When do corrections have an effect?Political Communication 35, 2 (2018), 196–219.
  9. Kathleen L Mosier, Linda J Skitka, Kristina J Korte, M Mouloua, and R Parasuraman. 1994. Cognitive and social psychological issues in flight crew/automation interaction. Human performance in automated systems: Current research and trends (1994), 191–197.
  10. Gordon Pennycook, Ziv Epstein, Mohsen Mosleh, Antonio A Arechar, Dean Eckles, and David G Rand. 2021. Shifting attention to accuracy can reduce misinformation online. Nature 592, 7855 (2021), 590–595.
  11. Gordon Pennycook, Jonathon McPhetres, Yunhao Zhang, Jackson G Lu, and David G Rand. 2020. Fighting COVID-19 misinformation on social media: Experimental evidence for a scalable accuracy-nudge intervention. Psychological science 31, 7 (2020), 770–780.
  12. Emily V Shaw, Mona Lynch, Sofia Laguna, and Steven J Frenda. 2021. Race, witness credibility, and jury deliberation in a simulated drug trafficking trial.Law and human behavior 45, 3 (2021), 215.
  13. Brent Smith and Greg Linden. 2017. Two decades of recommender systems at Amazon. com. Ieee internet computing 21, 3 (2017), 12–18.
  14. Sara Spray. 2016. Facebook Is Embroiled In A Row With Activists Over “Censorship”. Retrieved August 25, 2022 from https://www.buzzfeed.com/saraspary/facebook-in-dispute-with-pro-kurdish-activists-over-deleted

Cited by

See all 59

Metadata

Authors
Jahanbakhsh, F., Katsis, Y., Wang, D., Popa, L., & Muller, M.
Affiliations
  • MIT Computer Science and Artificial Intelligence Laboratory (US)
  • IBM Research - Almaden (US)
  • Northeastern University (US)
  • IBM (United States) (US)
Venue
Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems
Publisher
ACM
Published
2023-04-19
Pages
1-27
OpenAlex
W4366547816
Cited by
42
References
74
Citations by year
2026: 11 · 2025: 17 · 2024: 13 · 2023: 1
FWCI
39.00
Type
conference paper
Language
en
Open access
Locations
  • CHI Conference on Human Factors in Computing Systems (publishedVersion, cc-by)
  • DSpace@MIT (Massachusetts Institute of Technology) (submittedVersion, cc-by-nc)
Topics
  • Social Sciences › Sociology and Political Science › Misinformation and Its Impacts
  • Physical Sciences › Computer Science › Information Systems › Spam and Phishing Detection
  • Physical Sciences › Computer Science › Human-Computer Interaction › Innovative Human-Technology Interaction
Keywords
misinformation detection, personalized AI, social media, user feedback, trust in AI predictions, user study

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Exploring the Use of Personalized AI for Identifying Misinformation on Social Media. https://app.cochat.ai/papers/10.1145/3544548.3581219
The paper
Jahanbakhsh, F., Katsis, Y., Wang, D., Popa, L., & Muller, M. (2023). Exploring the Use of Personalized AI for Identifying Misinformation on Social Media. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3544548.3581219

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.