Philipp Schoenegger
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Philipp Schoenegger has a total of 511 co-authors. Here are the top co-authors by number of shared publications:
- Adrian Dominik Wojcik
Nicolaus Copernicus University in Toruń, Toruń, Poland 1
5 shared publications - Ahmed Skali
Department of Global Economics and Management, University of Groningen, Groningen, Netherlands
5 shared publications - Aleksandra Cichocka
School of Psychology University of Kent Kent UK
5 shared publications - Aleksandra Cislak
Faculty of Psychology SWPS University Warszawa Poland
5 shared publications - Alexander Cameron Walker
Department of Psychology, University of Waterloo, Waterloo, Canada
5 shared publications - Anat Perry
Psychology Department, The Hebrew University of Jerusalem, Jerusalem, Israel
5 shared publications
Latest Publications
- 18 May 2026
- Published by Elsevier in Journal of Development Economics
- Vol. 183
Forecasting long-run causal effects
- 18 May 2026
- Published by Elsevier in Journal of Development Economics
- Vol. 183
Forecasting long-run causal effects
- Open Access
- 12 February 2025
- Published by Springer Nature in Communications Psychology
- Vol. 3 (1) , 1-12
AI can outperform humans in predicting correlations between personality items
We assess the abilities of both specialized deep neural networks, such as PersonalityMap, and general LLMs, including GPT-4o and Claude 3 Opus, in understanding human personality by predicting correlations between personality questionnaire items. All AI models outperform the vast majority of laypeople and academic experts. However, we can improve the accuracy of individual correlation predictions by taking the median prediction per group to produce a “wisdom of the crowds” estimate. Thus, we also compare the median predictions from laypeople, academic experts, GPT-4o/Claude 3 Opus, and PersonalityMap. Based on medians, PersonalityMap and academic experts surpass both LLMs and laypeople on most measures. These results suggest that while advanced LLMs make superior predictions compared to most individual humans, specialized models like PersonalityMap can match even expert group-level performance in domain-specific tasks. This underscores the capabilities of large language models while emphasizing the continued relevance of specialized systems as well as human experts for personality research.
- 12 February 2025
- Published by Springer Nature in Communications Psychology
- Vol. 3 (1) , 1-12
Open AccessAI can outperform humans in predicting correlations between personality items
We assess the abilities of both specialized deep neural networks, such as PersonalityMap, and general LLMs, including GPT-4o and Claude 3 Opus, in understanding human personality by predicting correlations between personality questionnaire items. All AI models outperform the vast majority of laypeople and academic experts. However, we can improve the accuracy of individual correlation predictions by taking the median prediction per group to produce a “wisdom of the crowds” estimate. Thus, we also compare the median predictions from laypeople, academic experts, GPT-4o/Claude 3 Opus, and PersonalityMap. Based on medians, PersonalityMap and academic experts surpass both LLMs and laypeople on most measures. These results suggest that while advanced LLMs make superior predictions compared to most individual humans, specialized models like PersonalityMap can match even expert group-level performance in domain-specific tasks. This underscores the capabilities of large language models while emphasizing the continued relevance of specialized systems as well as human experts for personality research.
- Top Cited PapersOpen Access
- 10 February 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 15 (1) , 1-25
AI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracy
Large language models (LLMs) match and sometimes exceed human performance in many domains. This study explores the potential of LLMs to augment human judgment in a forecasting task. We evaluate the effect on human forecasters of two LLM assistants: one designed to provide high-quality (“superforecasting”) advice, and the other designed to be overconfident and base-rate neglecting, thus providing noisy forecasting advice. We compare participants using these assistants to a control group that received a less advanced model that did not provide numerical predictions or engage in explicit discussion of predictions. Participants ( N \(=\) 991) answered a set of six forecasting questions and had the option to consult their assigned LLM assistant throughout. Our preregistered analyses show that interacting with each of our frontier LLM assistants significantly enhances prediction accuracy by between 24% and 28% compared to the control group. Exploratory analyses showed a pronounced outlier effect in one forecasting item, without which we find that the superforecasting assistant increased accuracy by 41%, compared with 29% for the noisy assistant. We further examine whether LLM forecasting augmentation disproportionately benefits less skilled forecasters, degrades the wisdom-of-the-crowd by reducing prediction diversity, or varies in effectiveness with question difficulty. Our data do not consistently support these hypotheses. Our results suggest that access to a frontier LLM assistant, even a noisy one, can be a helpful decision aid in cognitively demanding tasks compared to a less powerful model that does not provide specific forecasting advice. However, the effects of outliers suggest that further research into the robustness of this pattern is needed.- 10 February 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Interactive Intelligent Systems
- Vol. 15 (1) , 1-25
Top Cited PapersOpen AccessAI-Augmented Predictions: LLM Assistants Improve Human Forecasting Accuracy
Large language models (LLMs) match and sometimes exceed human performance in many domains. This study explores the potential of LLMs to augment human judgment in a forecasting task. We evaluate the effect on human forecasters of two LLM assistants: one designed to provide high-quality (“superforecasting”) advice, and the other designed to be overconfident and base-rate neglecting, thus providing noisy forecasting advice. We compare participants using these assistants to a control group that received a less advanced model that did not provide numerical predictions or engage in explicit discussion of predictions. Participants ( N \(=\) 991) answered a set of six forecasting questions and had the option to consult their assigned LLM assistant throughout. Our preregistered analyses show that interacting with each of our frontier LLM assistants significantly enhances prediction accuracy by between 24% and 28% compared to the control group. Exploratory analyses showed a pronounced outlier effect in one forecasting item, without which we find that the superforecasting assistant increased accuracy by 41%, compared with 29% for the noisy assistant. We further examine whether LLM forecasting augmentation disproportionately benefits less skilled forecasters, degrades the wisdom-of-the-crowd by reducing prediction diversity, or varies in effectiveness with question difficulty. Our data do not consistently support these hypotheses. Our results suggest that access to a frontier LLM assistant, even a noisy one, can be a helpful decision aid in cognitively demanding tasks compared to a less powerful model that does not provide specific forecasting advice. However, the effects of outliers suggest that further research into the robustness of this pattern is needed. - Top Cited PapersOpen Access
- 8 November 2024
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 10 (45) , eadp1528
Wisdom of the silicon crowd: LLM ensemble prediction capabilities rival human crowd accuracy
Human forecasting accuracy improves through the “wisdom of the crowd” effect, in which aggregated predictions tend to outperform individual ones. Past research suggests that individual large language models (LLMs) tend to underperform compared to human crowd aggregates. We simulate a wisdom of the crowd effect with LLMs. Specifically, we use an ensemble of 12 LLMs to make probabilistic predictions about 31 binary questions, comparing them with those made by 925 human forecasters in a 3-month tournament. We show that the LLM crowd outperforms a no-information benchmark and is statistically indistinguishable from the human crowd. We also observe human-like biases, such as the acquiescence bias. In another study, we find that LLM predictions (of GPT-4 and Claude 2) improve when exposed to the median human prediction, increasing accuracy by 17 to 28%. However, simply averaging human and machine forecasts yields more accurate results. Our findings suggest that LLM predictions can rival the human crowd’s forecasting accuracy through simple aggregation.
- 8 November 2024
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 10 (45) , eadp1528
Top Cited PapersOpen AccessWisdom of the silicon crowd: LLM ensemble prediction capabilities rival human crowd accuracy
Human forecasting accuracy improves through the “wisdom of the crowd” effect, in which aggregated predictions tend to outperform individual ones. Past research suggests that individual large language models (LLMs) tend to underperform compared to human crowd aggregates. We simulate a wisdom of the crowd effect with LLMs. Specifically, we use an ensemble of 12 LLMs to make probabilistic predictions about 31 binary questions, comparing them with those made by 925 human forecasters in a 3-month tournament. We show that the LLM crowd outperforms a no-information benchmark and is statistically indistinguishable from the human crowd. We also observe human-like biases, such as the acquiescence bias. In another study, we find that LLM predictions (of GPT-4 and Claude 2) improve when exposed to the median human prediction, increasing accuracy by 17 to 28%. However, simply averaging human and machine forecasts yields more accurate results. Our findings suggest that LLM predictions can rival the human crowd’s forecasting accuracy through simple aggregation.
- Open Access
- 15 May 2024
- Published by Elsevier in Studies in History and Philosophy of Science
- Vol. 105, 85-98
Scientific realism, scientific practice, and science communication: An empirical investigation of academics and science communicators
- 15 May 2024
- Published by Elsevier in Studies in History and Philosophy of Science
- Vol. 105, 85-98
Open AccessScientific realism, scientific practice, and science communication: An empirical investigation of academics and science communicators
- Open Access
- 3 April 2024
- Published by Elsevier in Journal of Environmental Psychology
- Vol. 95
Moral hazards and solar radiation management: Evidence from a large-scale online experiment
- 3 April 2024
- Published by Elsevier in Journal of Environmental Psychology
- Vol. 95
Open AccessMoral hazards and solar radiation management: Evidence from a large-scale online experiment
- Top Cited PapersOpen Access
- 9 January 2024
- Published by Springer Nature in Behavior Research Methods
- Vol. 56 (6) , 5754-5770
Diminished diversity-of-thought in a standard large language model
We test whether large language models (LLMs) can be used to simulate human participants in social-science studies. To do this, we ran replications of 14 studies from the Many Labs 2 replication project with OpenAI’s text-davinci-003 model, colloquially known as GPT-3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the “correct answer” effect. Different runs of GPT-3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly “correct answer.” In one exploratory follow-up study, we found that a “correct answer” was robust to changing the demographic details that precede the prompt. In another, we found that most but not all “correct answers” were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT-3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported ‘GPT conservatives’ and ‘GPT liberals’ showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity of thought.
- 9 January 2024
- Published by Springer Nature in Behavior Research Methods
- Vol. 56 (6) , 5754-5770
Top Cited PapersOpen AccessDiminished diversity-of-thought in a standard large language model
We test whether large language models (LLMs) can be used to simulate human participants in social-science studies. To do this, we ran replications of 14 studies from the Many Labs 2 replication project with OpenAI’s text-davinci-003 model, colloquially known as GPT-3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the “correct answer” effect. Different runs of GPT-3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly “correct answer.” In one exploratory follow-up study, we found that a “correct answer” was robust to changing the demographic details that precede the prompt. In another, we found that most but not all “correct answers” were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT-3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported ‘GPT conservatives’ and ‘GPT liberals’ showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity of thought.
- Open Access
- 1 January 2024
- Published by Elsevier in SSRN Electronic Journal
Forecasting Long-Run Causal Effects
Randomized controlled trials (RCT) are the gold standard in causal inference and are often used by decision-makers in public and private sectors. However, RCTs are often expensive and take a long time to complete, making predicting its effects prior to completion practically important. In this paper, we test the forecasting accuracy of groups of experienced forecasters, academic experts, and lay people in predicting such long-run causal effects. To do this, we recruit 511 unique participants and collect 25,980 individual forecasts. Participants predict the short-run as well as long-run effects of seven RCTs that collected data on a treatment effect with a duration of at least 5 years. We find that experienced forecasters and academics outperform lay people. We also find evidence that experienced forecasters perform better than academics on our primary accuracy measure, and that this is, at least in part, due to their superior calibration. While we do find aggregation effects suggesting a wisdom of the crowds effect, we also show that neither experienced forecasters nor academics consistently outperform a set of simple benchmarks. Additionally, we also fail to find evidence that randomly assigned information about good forecasting practices, details about the RCT intervention, or the local study context improves forecasting accuracy. Overall, our finding that experienced forecasters can outperform academic experts is a novel addition to an otherwise mixed literature. However, the general pattern of results is largely in line with other work, showing that forecasting long-run events is extremely difficult and that improving forecasting performance remains somewhat intractable.
- 1 January 2024
- Published by Elsevier in SSRN Electronic Journal
Open AccessForecasting Long-Run Causal Effects
Randomized controlled trials (RCT) are the gold standard in causal inference and are often used by decision-makers in public and private sectors. However, RCTs are often expensive and take a long time to complete, making predicting its effects prior to completion practically important. In this paper, we test the forecasting accuracy of groups of experienced forecasters, academic experts, and lay people in predicting such long-run causal effects. To do this, we recruit 511 unique participants and collect 25,980 individual forecasts. Participants predict the short-run as well as long-run effects of seven RCTs that collected data on a treatment effect with a duration of at least 5 years. We find that experienced forecasters and academics outperform lay people. We also find evidence that experienced forecasters perform better than academics on our primary accuracy measure, and that this is, at least in part, due to their superior calibration. While we do find aggregation effects suggesting a wisdom of the crowds effect, we also show that neither experienced forecasters nor academics consistently outperform a set of simple benchmarks. Additionally, we also fail to find evidence that randomly assigned information about good forecasting practices, details about the RCT intervention, or the local study context improves forecasting accuracy. Overall, our finding that experienced forecasters can outperform academic experts is a novel addition to an otherwise mixed literature. However, the general pattern of results is largely in line with other work, showing that forecasting long-run events is extremely difficult and that improving forecasting performance remains somewhat intractable.
- Open Access
- 15 September 2023
- Published by Center for Open Science in PsyArXiv
Scientific Realism, Scientific Practice, and Science Communication. An Empirical Investigation of Academics and Science Communicators
We argue that the societal consequences of the scientific realism debate, in the context of science-to-public communication are often overlooked and careful theorizing about it needs further empirical groundwork. As such, we conducted a survey experiment with 130 academics (from physics, chemistry, and biology) and 137 science communicators. We provided them with a 11-item questionnaire probing their views of scientific realism and related concepts. Contra theoretical expectations, we find that (a) science communicators are generally more inclined towards scientific antirealism when compared to scientists in the same academic fields, though both groups show an inclination towards realism and (b) academics who engage in more theoretical work are not less (or more) realist than experimentalists. Lastly, (c), we fail to find differences with respect to selective realism but find that science communicators are significantly less epistemically voluntarist compared to their academic counterparts. Overall, our results provide first empirical evidence on the views of scientists and science communicators on scientific realism, with some results running contra to the theoretical expectations, opening up new empirical and theoretical research directions.
- 15 September 2023
- Published by Center for Open Science in PsyArXiv
Open AccessScientific Realism, Scientific Practice, and Science Communication. An Empirical Investigation of Academics and Science Communicators
We argue that the societal consequences of the scientific realism debate, in the context of science-to-public communication are often overlooked and careful theorizing about it needs further empirical groundwork. As such, we conducted a survey experiment with 130 academics (from physics, chemistry, and biology) and 137 science communicators. We provided them with a 11-item questionnaire probing their views of scientific realism and related concepts. Contra theoretical expectations, we find that (a) science communicators are generally more inclined towards scientific antirealism when compared to scientists in the same academic fields, though both groups show an inclination towards realism and (b) academics who engage in more theoretical work are not less (or more) realist than experimentalists. Lastly, (c), we fail to find differences with respect to selective realism but find that science communicators are significantly less epistemically voluntarist compared to their academic counterparts. Overall, our results provide first empirical evidence on the views of scientists and science communicators on scientific realism, with some results running contra to the theoretical expectations, opening up new empirical and theoretical research directions.
- Open Access
Diminished Diversity-of-Thought in a Standard Large Language Model
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.
Open AccessDiminished Diversity-of-Thought in a Standard Large Language Model
We test whether Large Language Models (LLMs) can be used to simulate human participants in social-science studies. To do this, we run replications of 14 studies from the Many Labs 2 replication project with OpenAI's text-davinci-003 model, colloquially known as GPT3.5. Based on our pre-registered analyses, we find that among the eight studies we could analyse, our GPT sample replicated 37.5% of the original results and 37.5% of the Many Labs 2 results. However, we were unable to analyse the remaining six studies due to an unexpected phenomenon we call the "correct answer" effect. Different runs of GPT3.5 answered nuanced questions probing political orientation, economic preference, judgement, and moral philosophy with zero or near-zero variation in responses: with the supposedly "correct answer." In one exploratory follow-up study, we found that a "correct answer" was robust to changing the demographic details that precede the prompt. In another, we found that most but not all "correct answers" were robust to changing the order of answer choices. One of our most striking findings occurred in our replication of the Moral Foundations Theory survey results, where we found GPT3.5 identifying as a political conservative in 99.6% of the cases, and as a liberal in 99.3% of the cases in the reverse-order condition. However, both self-reported 'GPT conservatives' and 'GPT liberals' showed right-leaning moral foundations. Our results cast doubts on the validity of using LLMs as a general replacement for human participants in the social sciences. Our results also raise concerns that a hypothetical AI-led future may be subject to a diminished diversity-of-thought.