ChatGPT Influence on Medical Decision-Making, Bias, and Equity: A Randomized Study of Clinicians Evaluating Clinical Vignettes.

In a randomized, pre-post intervention study, we evaluated the influence of a large language model (LLM) generative AI system on accuracy of physician decision-making and bias in healthcare. 50 US-licensed physicians reviewed a video clinical vignette, featuring actors representing different demographics (a White male or a Black female) with chest pain. Participants were asked to answer clinical questions around triage, risk, and treatment based on these vignettes, then asked to reconsider after receiving advice generated by ChatGPT+ (GPT4). The primary outcome was the accuracy of clinical decisions based on pre-established evidence-based guidelines. Results showed that physicians are willing to change their initial clinical impressions given AI assistance, and that this led to a significant improvement in clinical decision-making accuracy in a chest pain evaluation scenario without introducing or exacerbating existing race or gender biases. A survey of physician participants indicates that the majority expect LLM tools to play a significant role in clinical decision making.

Experimental evidence for structured information-sharing networks reducing medical errors.

Errors in clinical decision-making are disturbingly common. Recent studies have found that 10 to 15% of all clinical decisions regarding diagnoses and treatment are inaccurate. Here, we experimentally study the ability of structured information-sharing networks among clinicians to improve clinicians' diagnostic accuracy and treatment decisions. We use a pool of 2,941 practicing clinicians recruited from around the United States to conduct 84 independent group-level trials, ranging across seven different clinical vignettes for topics known to exhibit high rates of diagnostic or treatment error (e.g., acute cardiac events, geriatric care, low back pain, and diabetes-related cardiovascular illness prevention). We compare collective performance in structured information-sharing networks to collective performance in independent control groups, and find that networks significantly reduce clinical errors, and improve treatment recommendations, as compared to control groups of independent clinicians engaged in isolated reflection. Our results show that these improvements are not a result of simple regression to the group mean. Instead, we find that within structured information-sharing networks, the worst clinicians improved significantly while the best clinicians did not decrease in quality. These findings offer implications for the use of social network technologies to reduce errors among clinicians.

Scaling Up Change: A Critical Review and Practical Guide to Harnessing Social Norms for Climate Action.

Anthropogenic carbon emissions have the potential to trigger changes in climate and ecosystems that would be catastrophic for the well-being of humans and other species. Widespread shifts in production and consumption patterns are urgently needed to address climate change. Although transnational agreements and national policy are necessary for a transition to a fully decarbonized global economy, fluctuating political priorities and lobbying by vested interests have slowed these efforts. Against this backdrop, bottom-up pressure from social movements and shifting social norms may offer a complementary path to a more sustainable economy. Furthermore, norm change may be an important component of decarbonization policies by accelerating or strengthening the impacts of other demand-side measures. Individual actions and policy support are social processes-they are intimately linked to expectations about the actions and beliefs of others. Although prevailing social norms often reinforce the status quo and unsustainable development pathways, social dynamics can also create widespread and rapid shifts in cultural values and practices, including increasing pressure on politicians to enact ambitious policy. We synthesize literature on social-norm influence, measurement, and change from the perspectives of psychology, anthropology, sociology, and economics. We discuss the opportunities and challenges for the use of social-norm and social-tipping interventions to promote climate action. Social-norm interventions aimed at addressing climate change or other social dilemmas are promising but no panacea. They require in-depth contextual knowledge, ethical consideration, and situation-specific tailoring and testing to understand whether they can be effectively implemented at scale. Our review aims to provide practitioners with insights and tools to reflect on the promises and pitfalls of such interventions in diverse contexts.

The network science of collective intelligence.

In the last few years, breakthroughs in computational and experimental techniques have produced several key discoveries in the science of networks and human collective intelligence. This review presents the latest scientific findings from two key fields of research: collective problem-solving and the wisdom of the crowd. I demonstrate the core theoretical tensions separating these research traditions and show how recent findings offer a new synthesis for understanding how network dynamics alter collective intelligence, both positively and negatively. I conclude by highlighting current theoretical problems at the forefront of research on networked collective intelligence, as well as vital public policy challenges that require new research efforts.

The reduction of race and gender bias in clinical treatment recommendations using clinician peer networks in an experimental setting.

Bias in clinical practice, in particular in relation to race and gender, is a persistent cause of healthcare disparities. We investigated the potential of a peer-network approach to reduce bias in medical treatment decisions within an experimental setting. We created "egalitarian" information exchange networks among practicing clinicians who provided recommendations for the clinical management of patient scenarios, presented via standardized patient videos of actors portraying patients with cardiac chest pain. The videos, which were standardized for relevant clinical factors, presented either a white male actor or Black female actor of similar age, wearing the same attire and in the same clinical setting, portraying a patient with clinically significant chest pain symptoms. We found significant disparities in the treatment recommendations given to the white male patient-actor and Black female patient-actor, which when translated into real clinical scenarios would result in the Black female patient being significantly more likely to receive unsafe undertreatment, rather than the guideline-recommended treatment. In the experimental control group, clinicians who were asked to independently reflect on the standardized patient videos did not show any significant reduction in bias. However, clinicians who exchanged real-time information in structured peer networks significantly improved their clinical accuracy and showed no bias in their final recommendations. The findings indicate that clinician network interventions might be used in healthcare settings to reduce significant disparities in patient treatment.

Topological measures for identifying and predicting the spread of complex contagions.

The standard measure of distance in social networks - average shortest path length - assumes a model of "simple" contagion, in which people only need exposure to influence from one peer to adopt the contagion. However, many social phenomena are "complex" contagions, for which people need exposure to multiple peers before they adopt. Here, we show that the classical measure of path length fails to define network connectedness and node centrality for complex contagions. Centrality measures and seeding strategies based on the classical definition of path length frequently misidentify the network features that are most effective for spreading complex contagions. To address these issues, we derive measures of complex path length and complex centrality, which significantly improve the capacity to identify the network structures and central individuals best suited for spreading complex contagions. We validate our theory using empirical data on the spread of a microfinance program in 43 rural Indian villages.

Experimental evidence for scale-induced category convergence across populations.

Individuals vary widely in how they categorize novel and ambiguous phenomena. This individual variation has led influential theories in cognitive and social science to suggest that communication in large social groups introduces path dependence in category formation, which is expected to lead separate populations toward divergent cultural trajectories. Yet, anthropological data indicates that large, independent societies consistently arrive at highly similar category systems across a range of topics. How is it possible for diverse populations, consisting of individuals with significant variation in how they categorize the world, to independently construct similar category systems? Here, we investigate this puzzle experimentally by creating an online "Grouping Game" in which we observe how people in small and large populations collaboratively construct category systems for a continuum of ambiguous stimuli. We find that solitary individuals and small groups produce highly divergent category systems; however, across independent trials with unique participants, large populations consistently converge on highly similar category systems. A formal model of critical mass dynamics in social networks accurately predicts this process of scale-induced category convergence. Our findings show how large communication networks can filter lexical diversity among individuals to produce replicable society-level patterns, yielding unexpected implications for cultural evolution.

Impact of network structure on collective learning: An experimental study in a data science competition.

Do efficient communication networks accelerate solution discovery? The most prominent theory of organizational design for collective learning maintains that informationally efficient collaboration networks increase a group's ability to find innovative solutions to complex problems. We test this idea against a competing theory that argues that communication networks that are less efficient for information transfer will increase the discovery of novel solutions to complex problems. We conducted a series of experimentally designed Data Science Competitions, in which we manipulated the efficiency of the communication networks among distributed groups of data scientists attempting to find better solutions for complex statistical modeling problems. We present findings from 16 independent competitions, where individuals conduct greedy search and only adopt better solutions. We show that groups with inefficient communication networks consistently discovered better solutions. In every experimental trial, groups with inefficient networks outperformed groups with efficient networks, as measured by both the group's average solution quality and the best solution found by a group member.

Networked collective intelligence improves dissemination of scientific information regarding smoking risks.

Despite substantial investments in public health campaigns, misunderstanding of health-related scientific information is pervasive. This is especially true in the case of tobacco use, where smokers have been found to systematically misperceive scientific information about the negative health effects of smoking, in some cases leading smokers to increase their pro-smoking bias. Here, we extend recent work on 'networked collective intelligence' by testing the hypothesis that allowing smokers and nonsmokers to collaboratively evaluate anti-smoking advertisements in online social networks can improve their ability to accurately assess the negative health effects of tobacco use. Using Amazon's Mechanical Turk, we conducted an online experiment where smokers and nonsmokers (N = 1600) were exposed to anti-smoking advertisements and asked to estimate the negative health effects of tobacco use, either on their own or in the presence of peer influence in a social network. Contrary to popular predictions, we find that both smokers and nonsmokers were surprisingly inaccurate at interpreting anti-smoking messages, and their errors persisted if they continued to interpret these messages on their own. However, smokers and nonsmokers significantly improved in their ability to accurately interpret anti-smoking messages by sharing their opinions in structured online social networks. Specifically, subjects in social networks reduced the error of their risk estimates by over 10 times more than subjects who revised solely based on individual reflection (p < 0.001, 10 experimental trials in total). These results suggest that social media networks may be used to activate social learning that improves the public's ability to accurately interpret vital public health information.

Birth Control Connect: A randomized trial of an online group to disseminate contraceptive information.

OBJECTIVE: We sought to test whether participation in an online group including IUD users influenced IUD-related knowledge, attitudes, and behavior among IUD non-users, as a proof-of-concept evaluation of information dissemination for less commonly used or novel contraceptives. STUDY DESIGN: We conducted a blinded, randomized controlled trial on the effect of online communication with IUD users within an online program called Birth Control Connect. Participants were women age 18-45 living in the United States who had never used an IUD. We invited participants randomized to the intervention to join two-week, nine-member discussion groups including four satisfied IUD users and five IUD non-users; we invited control participants to groups including nine IUD non-users. We performed chi-squared tests on IUD knowledge, information-seeking, informational support and use in immediate post-surveys, and t-tests comparing change in IUD attitudes and frequency of logins to discussion groups. RESULTS: We invited 488 IUD non-users and enrolled them into 70 groups between October 2015 and April 2016. We found increased positive attitudes towards the IUD in the intervention arm (0.65-point increase between pre- and post-surveys, versus 0.05 mean change for control arm, p = 0.03 for hormonal IUD, with a trend in the same direction for the non-hormonal IUD). Informational support also increased, with 70.3% of intervention arm participants self-reporting that they gained a better idea of what the IUD would be like, compared to 51.3% in control arm (p < 0.01). Of intervention participants, 63.3% versus 51.3% of control participants reported gaining new information from their group (p = 0.03). There were no differences in correct responses to knowledge items or information-seeking between groups. CONCLUSIONS: Online exposure to IUD users increased positive attitudes toward the IUD and informational support for decision-making about the IUD among non-users. IMPLICATIONS STATEMENT: Online spaces provide a promising environment for the exchange of accurate, useful contraceptive information based on real user experiences. Interventions aiming to harness social communication through structured online conversations (e.g., on existing social media platforms) about user experiences with lesser-known contraceptive methods such as the IUD may be worthwhile.

Facts or stories? How to use social media for cervical cancer prevention: A multi-method study of the effects of sender type and content type on increased message sharing.

Social media has become a valuable tool for disseminating cancer prevention information. However, the design of messages for achieving wide dissemination remains poorly understood. We conducted a multi-method study to identify the effects of sender type (individuals or organizations) and content type (personal experiences or factual information) on promoting the spread of cervical cancer prevention messages over social media. First, we used observational Twitter data to examine correlations between sender type and content type with retweet activity. Then, to confirm the causal impact of message properties, we constructed 900 experimental tweets according to a 2 (sender type) by 2 (content type) factorial design and tested their probabilities of being shared in an online platform. A total of 782 female participants were randomly assigned to 87 independent 9-person online groups and each received a unique message feed of 100 tweets drawn from the 4 experimental cells over 5 days. We conducted both tweet-level and group-level analyses to examine the causal effects of tweet properties on influencing sharing behaviors. Personal experience tweets and organizational senders were associated with more retweets. However, the experimental study revealed that informational tweets were shared significantly more (19%, 95% CI: 11 to 27) than personal experience tweets; and organizational senders were shared significantly more (10%, 95% CI: 3 to 18) than individual senders. While rare personal experience messages can achieve large success, they are generally unsuccessful; however, there is a reproducible causal effect of messages that use organizational senders and factual information for achieving greater peer-to-peer dissemination.

The wisdom of partisan crowds.

Theories in favor of deliberative democracy are based on the premise that social information processing can improve group beliefs. While research on the "wisdom of crowds" has found that information exchange can increase belief accuracy on noncontroversial factual matters, theories of political polarization imply that groups will become more extreme-and less accurate-when beliefs are motivated by partisan political bias. A primary concern is that partisan biases are associated not only with more extreme beliefs, but also with a diminished response to social information. While bipartisan networks containing both Democrats and Republicans are expected to promote accurate belief formation, politically homogeneous networks are expected to amplify partisan bias and reduce belief accuracy. To test whether the wisdom of crowds is robust to partisan bias, we conducted two web-based experiments in which individuals answered factual questions known to elicit partisan bias before and after observing the estimates of peers in a politically homogeneous social network. In contrast to polarization theories, we found that social information exchange in homogeneous networks not only increased accuracy but also reduced polarization. Our results help generalize collective intelligence research to political domains.

Perceptions of cervical cancer prevention on Twitter uncovered by different sampling strategies.

INTRODUCTION: Cervical cancer prevention is possible through use of the HPV vaccine and Pap tests, yet the vaccine remains underutilized. METHODS: We obtained publicly-available Twitter data from 2014 using three sampling strategies (top-ranked, simple random sample, and topic model) based on key words related to cervical cancer prevention. We conducted a content analysis of 100 tweets from each of the three samples and examined the extent to which the narratives and frequency of themes differed across samples. RESULTS: Advocacy-related tweets constituted the most prevalent theme to emerge across all three sample types, and were most frequently found in the top-ranked sample. A random sample detected the same themes as topic modeling, but the relative frequency of themes identified from topic modeling fell in-between top-ranked and random samples. DISCUSSION: Variations in themes uncovered by different sampling methods suggest it is useful to qualitatively assess the relative frequency of themes to better understand the breadth and depth of social media conversations about health. CONCLUSIONS: Future studies using social media data should consider sampling methods to uncover a wider breadth of conversations about health on social media.

Social Media as a Tool to Promote Health Awareness: Results from an Online Cervical Cancer Prevention Study.

Online social media platforms represent a promising opportunity for public health promotion. Research is limited, however, on the effectiveness of social media at improving knowledge and awareness of health topics and motivating healthy behavior change. Therefore, we investigated whether participation in an online social media platform and receipt of brief, tailored messages is effective at increasing knowledge, awareness, and prevention behaviors related to human papillomavirus (HPV) and cervical cancer. We conducted an online study in which 782 recruited participants were consecutively assigned to nine-person groups on a social media platform. Participants were shown a unique random set of 20 tailored messages per day over five days. Participants completed a baseline and post survey to assess their knowledge, awareness, and prevention behaviors related to HPV and cervical cancer. There were no statistically significant changes in knowledge and prevention behaviors from the baseline to the post survey among study participants. There was a modest, statistically significant change in response to whether participants had ever heard of HPV, increasing from 90 to 94% (p = 0.003). Our findings suggest that most study participants had substantial knowledge, awareness, and engagement in positive behaviors related to cervical cancer prevention at the start of the study. Nevertheless, we found that HPV awareness can be increased through brief participation in an online social media platform and receipt of tailored health messages. Further investigation that explores how social media can be used to improve knowledge and adoption of healthy behaviors related to cervical cancer is warranted.

Social learning and partisan bias in the interpretation of climate trends.

Vital scientific communications are frequently misinterpreted by the lay public as a result of motivated reasoning, where people misconstrue data to fit their political and psychological biases. In the case of climate change, some people have been found to systematically misinterpret climate data in ways that conflict with the intended message of climate scientists. While prior studies have attempted to reduce motivated reasoning through bipartisan communication networks, these networks have also been found to exacerbate bias. Popular theories hold that bipartisan networks amplify bias by exposing people to opposing beliefs. These theories are in tension with collective intelligence research, which shows that exchanging beliefs in social networks can facilitate social learning, thereby improving individual and group judgments. However, prior experiments in collective intelligence have relied almost exclusively on neutral questions that do not engage motivated reasoning. Using Amazon's Mechanical Turk, we conducted an online experiment to test how bipartisan social networks can influence subjects' interpretation of climate communications from NASA. Here, we show that exposure to opposing beliefs in structured bipartisan social networks substantially improved the accuracy of judgments among both conservatives and liberals, eliminating belief polarization. However, we also find that social learning can be reduced, and belief polarization maintained, as a result of partisan priming. We find that increasing the salience of partisanship during communication, both through exposure to the logos of political parties and through exposure to the political identities of network peers, can significantly reduce social learning.

Experimental evidence for tipping points in social convention.

Theoretical models of critical mass have shown how minority groups can initiate social change dynamics in the emergence of new social conventions. Here, we study an artificial system of social conventions in which human subjects interact to establish a new coordination equilibrium. The findings provide direct empirical demonstration of the existence of a tipping point in the dynamics of changing social conventions. When minority groups reached the critical mass-that is, the critical group size for initiating social change-they were consistently able to overturn the established behavior. The size of the required critical mass is expected to vary based on theoretically identifiable features of a social setting. Our results show that the theoretically predicted dynamics of critical mass do in fact emerge as expected within an empirical system of social coordination.