arXiv subjects

Abdullah Almaatouq

Publications and source records attributed to Abdullah Almaatouq.

15 recordsLinked to original sources

arXiv2026

Post-training makes large language models less human-like

Large language models (LLMs) are increasingly used as surrogates for human participants, but it remains unclear which models best capture human behavior and why. To address this, we introduce Psych-201, a novel dataset that enables us to measure behavioral alignment at scale. We find that post-training -- the stage that turns base models into useful assistants -- consistently reduces alignment with human behavior across model families, sizes, and objectives. Moreover, this misalignment widens in newer model generations even as base models continue to improve. Finally, we find that persona-induction -- a popular technique for eliciting human-like behavior by conditioning models on participant-specific information -- does not improve predictions at the level of individuals. Taken together, our results suggest that the very processes that are currently employed to turn LLMs into useful assistants also make them less accurate models of human behavior.

cs.CL↗

arXiv2026

Evaluating Human-AI Safety: A Framework for Measuring Harmful Capability Uplift

Current frontier AI safety evaluations emphasize static benchmarks, third-party annotations, and red-teaming. In this position paper, we argue that AI safety research should focus on human-centered evaluations that measure harmful capability uplift: the marginal increase in a user's ability to cause harm with a frontier model beyond what conventional tools already enable. We frame harmful capability uplift as a core AI safety metric, ground it in prior social science research, and provide concrete methodological guidance for systematic measurement. We conclude with actionable steps for developers, researchers, funders, and regulators to make harmful capability uplift evaluation a standard practice.

cs.CY↗

arXiv2025

Integrative Experiments Identify How Punishment Impacts Welfare in Public Goods Games

Punishment as a mechanism for promoting cooperation has been studied extensively for more than two decades, but its effectiveness remains a matter of dispute. Here, we examine how punishment's impact varies across cooperative settings through a large-scale integrative experiment. We vary 14 parameters that characterize public goods games, sampling 360 experimental conditions and collecting 147,618 decisions from 7,100 participants. Our results reveal striking heterogeneity in punishment effectiveness: while punishment consistently increases contributions, its impact on payoffs (i.e., efficiency) ranges from dramatically enhancing welfare (up to 43% improvement) to severely undermining it (up to 44% reduction) depending on the cooperative context. To characterize these patterns, we developed models that outperformed human forecasters (laypeople and domain experts) in predicting punishment outcomes in new experiments. Communication emerged as the most predictive feature, followed by contribution framing (opt-out vs. opt-in), contribution type (variable vs. all-or-nothing), game length (number of rounds), peer outcome visibility (whether participants can see others' earnings), and the availability of a reward mechanism. Interestingly, however, most of these features interact to influence punishment effectiveness rather than operating independently. For example, the extent to which longer games increase the effectiveness of punishment depends on whether groups can communicate. Together, our results refocus the debate over punishment from whether or not it "works" to the specific conditions under which it does and does not work. More broadly, our study demonstrates how integrative experiments can be combined with machine learning to uncover generalizable patterns, potentially involving interactions between multiple features, and help generate novel explanations in complex social phenomena.

econ.GN↗

arXiv2024

When combinations of humans and AI are useful: A systematic review and meta-analysis

Inspired by the increasing use of AI to augment humans, researchers have studied human-AI systems involving different tasks, systems, and populations. Despite such a large body of work, we lack a broad conceptual understanding of when combinations of humans and AI are better than either alone. Here, we addressed this question by conducting a meta-analysis of over 100 recent experimental studies reporting over 300 effect sizes. First, we found that, on average, human-AI combinations performed significantly worse than the best of humans or AI alone. Second, we found performance losses in tasks that involved making decisions and significantly greater gains in tasks that involved creating content. Finally, when humans outperformed AI alone, we found performance gains in the combination, but when the AI outperformed humans alone we found losses. These findings highlight the heterogeneity of the effects of human-AI collaboration and point to promising avenues for improving human-AI systems.

cs.HC↗

arXiv2022

A Test for Evaluating Performance in Human-Computer Systems

The Turing test for comparing computer performance to that of humans is well known, but, surprisingly, there is no widely used test for comparing how much better human-computer systems perform relative to humans alone, computers alone, or other baselines. Here, we show how to perform such a test using the ratio of means as a measure of effect size. Then we demonstrate the use of this test in three ways. First, in an analysis of 79 recently published experimental results, we find that, surprisingly, over half of the studies find a decrease in performance, the mean and median ratios of performance improvement are both approximately 1 (corresponding to no improvement at all), and the maximum ratio is 1.36 (a 36% improvement). Second, we experimentally investigate whether a higher performance improvement ratio is obtained when 100 human programmers generate software using GPT-3, a massive, state-of-the-art AI system. In this case, we find a speed improvement ratio of 1.27 (a 27% improvement). Finally, we find that 50 human non-programmers using GPT-3 can perform the task about as well as--and less expensively than--the human programmers. In this case, neither the non-programmers nor the computer would have been able to perform the task alone, so this is an example of a very strong form of human-computer synergy.

cs.HC↗

arXiv2021

Will We Trust What We Don't Understand? Impact of Model Interpretability and Outcome Feedback on Trust in AI

Despite AI's superhuman performance in a variety of domains, humans are often unwilling to adopt AI systems. The lack of interpretability inherent in many modern AI techniques is believed to be hurting their adoption, as users may not trust systems whose decision processes they do not understand. We investigate this proposition with a novel experiment in which we use an interactive prediction task to analyze the impact of interpretability and outcome feedback on trust in AI and on human performance in AI-assisted prediction tasks. We find that interpretability led to no robust improvements in trust, while outcome feedback had a significantly greater and more reliable effect. However, both factors had modest effects on participants' task performance. Our findings suggest that (1) factors receiving significant attention, such as interpretability, may be less effective at increasing trust than factors like outcome feedback, and (2) augmenting human performance via AI systems may not be a simple matter of increasing trust in AI, as increased trust is not always associated with equally sizable improvements in performance. These findings invite the research community to focus not only on methods for generating interpretations but also on techniques for ensuring that interpretations impact trust and performance in practice.

cs.AI↗

arXiv2021

When social influence promotes the wisdom of crowds

Whether, and under what conditions, groups exhibit "crowd wisdom" has been a major focus of research across the social and computational sciences. Much of this work has focused on the role of social influence in promoting the wisdom of the crowd versus leading the crowd astray, resulting in conflicting conclusions about how the social network structure determines the impact of social influence. Here, we demonstrate that it is not enough to consider the network structure in isolation. Using theoretical analysis, numerical simulation, and reanalysis of four experimental datasets (totaling 2,885 human subjects), we find that the wisdom of crowds critically depends on the interaction between (i) the centralization of the social influence network and (ii) the distribution of the initial, individual estimates. By adopting a framework that integrates both the structure of the social influence and the distribution of the initial estimates, we bring previously conflicting results under one theoretical framework and clarify the effects of social influence on the wisdom of crowds.

cs.SI↗

arXiv2021

Network Structures of Collective Intelligence: The Contingent Benefits of Group Discussion

Research on belief formation has produced contradictory findings on whether and when communication between group members will improve the accuracy of numeric estimates such as economic forecasts, medical diagnoses, and job candidate assessments. While some evidence suggests that carefully mediated processes such as the "Delphi method" produce more accurate beliefs than unstructured discussion, others argue that unstructured discussion outperforms mediated processes. Still others argue that independent individuals produce the most accurate beliefs. This paper shows how network theories of belief formation can resolve these inconsistencies, even when groups lack apparent structure as in informal conversation. Emergent network structures of influence interact with the pre-discussion belief distribution to moderate the effect of communication on belief formation. As a result, communication sometimes increases and sometimes decreases the accuracy of the average belief in a group. The effects differ for mediated processes and unstructured communication, such that the relative benefit of each communication format depends on both group dynamics as well as the statistical properties of pre-interaction beliefs. These results resolve contradictions in previous research and offer practical recommendations for teams and organizations.

econ.GN↗

arXiv2020

Empirica: a virtual lab for high-throughput macro-level experiments

Virtual labs allow researchers to design high-throughput and macro-level experiments that are not feasible in traditional in-person physical lab settings. Despite the increasing popularity of online research, researchers still face many technical and logistical barriers when designing and deploying virtual lab experiments. While several platforms exist to facilitate the development of virtual lab experiments, they typically present researchers with a stark trade-off between usability and functionality. We introduce Empirica: a modular virtual lab that offers a solution to the usability-functionality trade-off by employing a "flexible defaults" design strategy. This strategy enables us to maintain complete "build anything" flexibility while offering a development platform that is accessible to novice programmers. Empirica's architecture is designed to allow for parameterizable experimental designs, reusable protocols, and rapid development. These features will increase the accessibility of virtual lab experiments, remove barriers to innovation in experiment design, and enable rapid progress in the understanding of distributed human computation.

cs.HC↗

arXiv2020

The Wisdom of the Network: How Adaptive Networks Promote Collective Intelligence

Social networks continuously change as new ties are created and existing ones fade. It is widely noted that our social embedding exerts a strong influence on what information we receive and how we form beliefs and make decisions. However, most empirical studies on the role of social networks in collective intelligence have overlooked the dynamic nature of social networks and its role in fostering adaptive collective intelligence. It remains unknown (1) how network structures adapt to the attributes of individuals, and (2) whether this adaptation promotes the accuracy of individual and collective decisions. Here, we answer these questions through a series of behavioral experiments and supporting simulations. Our results reveal that social network plasticity in the presence of feedback, can adapt to biased and changing information environments, and produce collective estimates that are more accurate than their best-performing member. We explore two mechanisms that explain these results: (1) a global adaptation mechanism where the structural connectivity of the network itself changes such that it amplifies the estimates of high-performing members within the group; (2) a local adaptation mechanism where accurate individuals are more resistant to social influence, and therefore their initial belief is weighted in the collective estimate disproportionately. Thereby, our findings substantiate the role of social network plasticity and feedback as adaptive mechanisms for refining individual and collective judgments.

cs.SI↗

arXiv2018

Winning Models for GPA, Grit, and Layoff in the Fragile Families Challenge

In this paper, we discuss and analyze our approach to the Fragile Families Challenge. The challenge involved predicting six outcomes for 4,242 children from disadvantaged families from around the United States. The data consisted of over 12,000 features (covariates) about the children and their parents, schools, and overall environments from birth to age 9. Our approach relied primarily on existing data science techniques, including: (1) data preprocessing: elimination of low variance features, imputation of missing data, and construction of composite features; (2) feature selection through univariate Mutual Information and extraction of non-zero LASSO coefficients; (3) three machine learning models: Random Forest, Elastic Net, and Gradient-Boosted Trees; and finally (4) prediction aggregation according to performance. The top-performing submissions produced winning out-of-sample predictions for three outcomes: GPA, grit, and layoff. However, predictions were at most 20% better than a baseline that predicted the mean value of the training data of each outcome.

stat.AP↗

arXiv2018

Dynamical Networks of Influence in Small Group Discussions

In many domains of life, business and management, numerous problems are addressed by small groups of individuals engaged in face-to-face discussions. While research in social psychology has a long history of studying the determinants of small group performances, the internal dynamics that govern a group discussion is not yet well understood. Here, we rely on computational methods based on network analyses and opinion dynamics to described how individuals influence each other during a group discussion. We consider the situation in which a small group of three individuals engages in a discussion to solve an estimation task. We propose a model describing how group members gradually influence each other and revise their judgments over the course of the discussion. The main component of the model is an influence network - a weighted, directed graph that determines the extent to which individuals influence each other during the discussion. In simulations, we first study the optimal structure of the influence network that yields the best group performances. Then, we implement a social learning process by which individuals adapt to the past performance of their peers, thereby affecting the structure of the influence network in the long run. We explore the mechanisms underlying the emergence of efficient or maladaptive networks and show that the influence network can converge towards the optimal one, but only when individuals exhibit a social discounting bias by downgrading the relative performances of their peers. Finally, we find a late-speaker effect, whereby individuals who speak later in the discussion are perceived more positively in the long run and are thus more influential. The numerous predictions of the model can serve as a basis for future experiments, and this work opens research on small group discussion to computational social sciences.

physics.soc-ph↗

arXiv2016

The Role of Reciprocity and Directionality of Friendship Ties in Promoting Behavioral Change

Friendship is a fundamental characteristic of human beings and usually assumed to be reciprocal in nature. Despite this common expectation, in reality, not all friendships by default are reciprocal nor created equal. Here, we show that reciprocated friendships are more intimate and they are substantially different from those that are not. We examine the role of reciprocal ties in inducing more effective peer pressure in a cooperative arrangements setting and find that the directionality of friendship ties can significantly limit the ability to persuade others to act. Specifically, we observe a higher behavioral change and more effective peer-influence when subjects shared reciprocal ties with their peers compared to sharing unilateral ones. Moreover, through spreading process simulation, we find that although unilateral ties diffuse behaviors across communities, reciprocal ties play more important role at the early stages of the diffusion process.

cs.SI↗

arXiv2016

Mobile Communication Signatures of Unemployment

The mapping of populations socio-economic well-being is highly constrained by the logistics of censuses and surveys. Consequently, spatially detailed changes across scales of days, weeks, or months, or even year to year, are difficult to assess; thus the speed of which policies can be designed and evaluated is limited. However, recent studies have shown the value of mobile phone data as an enabling methodology for demographic modeling and measurement. In this work, we investigate whether indicators extracted from mobile phone usage can reveal information about the socio-economical status of microregions such as districts (i.e., average spatial resolution < 2.7km). For this we examine anonymized mobile phone metadata combined with beneficiaries records from unemployment benefit program. We find that aggregated activity, social, and mobility patterns strongly correlate with unemployment. Furthermore, we construct a simple model to produce accurate reconstruction of district level unemployment from their mobile communication patterns alone. Our results suggest that reliable and cost-effective economical indicators could be built based on passively collected and anonymized mobile phone data. With similar data being collected every day by telecommunication services across the world, survey-based methods of measuring community socioeconomic status could potentially be augmented or replaced by such passive sensing methods in the future.

cs.SI↗

arXiv2016

Complex Systems and a Computational Social Science Perspective on the Labor Market

Labor market institutions are central for modern economies, and their polices can directly affect unemployment rates and economic growth. At the individual level, unemployment often has a detrimental impact on people's well-being and health. At the national level, high employment is one of the central goals of any economic policy, due to its close association with national prosperity. The main goal of this thesis is to highlight the need for frameworks that take into account the complex structure of labor market interactions. In particular, we explore the benefits of leveraging tools from computational social science, network science, and data-driven theories to measure the flow of opportunities and information in the context of the labor market. First, we investigate our key hypothesis, which is that opportunity/information flow through weak ties, and this is a key determinant of the length of unemployment. We then extend the idea of opportunity/information flow to clusters of other economic activities, where we expect the flow within clusters of related activities to be higher than within isolated activities. This captures the intuition that within related activities there are more "capitals" involved and that such activities require similar "capabilities." Therefore, more extensive clusters of economic activities should generate greater growth through exploiting the greater flow of opportunities and information. We quantify the opportunity/information flow using a complexity measure of two economic activities (i.e. jobs and exports).

cs.SI↗