Social preferences toward humans and machines: a systematic experiment on the role of machine payoffs
Open accessAbstract
There is growing interest in the field of cooperative artificial intelligence (AI), that is, settings in which humans and machines cooperate. By now, more than 160 studies from various disciplines have reported on how people cooperate with machines in behavioral experiments. Our systematic review of the experimental instructions reveals that the implementation of the machine payoffs and the information participants receive about them differ drastically across these studies. In an online experiment ( N = 1,198), we compare how these different payoff implementations shape people’s revealed social preferences toward machines. When matched with machine partners, people reveal substantially stronger social preferences and reciprocity when they know that a human beneficiary receives the machine payoffs than when they know that no such “human behind the machine” exists. When participants are not informed about machine payoffs, we found weak social preferences toward machines. Comparing survey answers with those from a follow-up study ( N = 150), we conclude that people form their beliefs about machine payoffs in a self-serving way. Thus, our results suggest that the extent to which humans cooperate with machines depends on the implementation and information about the machine’s earnings.
AI snapshot
Written from the paper’s full text.
TL;DR
In an online experiment (N = 1,198; 899 decision makers), people showed stronger social preferences and reciprocity toward machine partners when a human beneficiary received the machine’s payoffs than when no human beneficiary existed. When no payoff information was provided, social preferences were weak, and follow-up evidence suggests people form self-serving beliefs about machine payoffs.
Key findings
- Sharing rates were highest toward fellow humans (29.93%) and lower toward machines with a human beneficiary (token player: 24.31%; programmer: 23.51%), and lowest toward machines without a human beneficiary (machine-earns: 13.42%; nobody-earns: 9.36%) or with no information (14.47%); no-information was statistically indistinguishable from machine-earns (p = .984).
- Behindness aversion (α) was stable around 0 across machine treatments (−0.12 ≤ α ≤ 0.04), but aheadness aversion (β) was present when a human earned the machine payoffs (token player: β = 0.24; programmer: β = 0.25) and dropped to 0 when no human beneficiary existed or no information was given (−0.01 ≤ β ≤ 0.05; pairwise ps < .001, Cohen’s ds > 0.77).
- Reciprocity toward machines with a human beneficiary resembled human–human reciprocity, but disappeared without a human beneficiary or payoff information (−0.06 ≤ γ ≤ 0.007; −0.06 ≤ δ ≤ −0.03; positive-reciprocity comparisons ps < .001, Cohen’s ds > 0.71).
- In the no-information treatment, 41.9% believed the money was not paid out, versus 18.18% in a follow-up study of uninvolved participants (χ²(1) = 16.34, p < .001); belief that the experimenter kept the money was 3.81% vs. 21.42% (χ²(1) = 14.42, p < .001).
Methods
- The paper first systematically categorized machine-payoff implementations in 160 behavioral studies, then ran a main online experiment with U.S. Prolific participants (total N = 1,198; 899 decision makers; age M = 35.75 years, SD = 10.91; 54.73% female) across six between-subjects treatments: fellow-human, programmer-behind-machine, token-player-behind-machine, machine-earns, nobody-earns, and no-information.
- Participants played a dictator game and a reciprocity game with 30 binary choices each, one randomly paid; the reciprocity game used the strategy method. Machine players were programmed to imitate human first-mover behavior using data from fellow-human treatments.
- Revealed social preferences were estimated with a random-utility model extending Charness and Rabin (2002) and Bellemare et al. (2011), yielding parameters θ = (α, β, γ, δ) via maximum likelihood; a finite mixture model identified preference types. A follow-up study (N = 150) had participants read the same no-information instructions without playing the incentivized games and report beliefs about machine payoffs.
Limitations
- Paper-stated: the study focused on one-shot dyadic settings; future research is needed on repeated interactions and one-to-many interactions. The paper also notes an open question whether the lack of deservingness differences stems from machine intermediaries or platform-based in-group favoritism, and suggests future within-subjects designs to study intraindividual variance.
- Evident from the text: the main study’s belief measure was collected after incentivized decisions, so the self-serving-belief interpretation rests on comparing with a follow-up sample that did not make incentivized decisions; the paper does not frame this as a limitation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
0 of 37 cited sources checked · 0 supported · 0 flagged
- Findingown data
The systematic review of experimental instructions reveals that the implementation of machine payoffs and the information participants receive about them differ drastically across studies.
“Our systematic review of the experimental instructions reveals that the implementation of the machine payoffs and the information participants receive about them differ drastically across these studies.”
- Also stated in: § Different Implementations of Machine Players · plain
“Existing research designs diverge regarding the crucial aspects of (a) what happens to the payoffs earned by algorithms and (b) what information about the payoffs is provided to the participants.”
Quote in the paperSubstantial claim - Also stated in: § Different Implementations of Machine Players · plain
- Findingown data
Cohen's d effect sizes for the comparison of fellow-human with machine-earns, nobody-earns, and no-information treatments were > 1 in absolute value.
“and > 1 for the comparison with the machine-earns (13.42%), nobody-earns (9.36%) and no-information (14.47%) treatments.”
Quote in the paperNot a substantial claim - Findingown data
The proportion of participants sharing is significantly higher when a human behind the machine (token player or programmer) earned the machine's payoffs than when no human behind the machine earned the payoffs (machine-earns and nobody-earns treatments), all pairwise comparisons p < .01 (Cohen's ds > 1.1 in absolute value).
“The proportion of participants sharing is significantly higher when a human behind the machine (token player or programmer) earned the machine’s payoffs than when no human behind the machine earned the payoffs (machine-earns and nobody-earns treatments), all pairwise comparisons p < .01 (Cohen’s ds > 1.1 in absolute value).”
Quote in the paperSubstantial claim - Findingcitation
In 14 experiments, participants learned that the machine earns the payoffs.
“Third, in 14 experiments, participants learned that the machine earns the payoffs.”
Quote in the paperNot a substantial claim - Findingown data
Only a very small fraction of participants (3%-5%) chose the nonselfish option when there was no human beneficiary and Option Y was not efficient and assigned strictly higher payoffs to the receiver than Option X.
“The data revealed that only a very small fraction of participants (3%–5%) chose the nonselfish option when there was no human beneficiary and that Option Y was not efficient and assigned strictly higher payoffs to the receiver than Option X.”
Quote in the paperSubstantial claim
- Findingown data
Sharing rates by participants who received no information are statistically indistinguishable from the machine-earns treatment (p = .984).
“Moreover, sharing rates by participants who received no information are statistically indistinguishable from the machine-earns treatment (p = .984).”
Quote in the paperSubstantial claim - Findingcitation
In the fellow-human treatment, the estimate of α is relatively small (α = −0.07), albeit slightly negative.
“Our results in the fellow-human treatment largely replicate previous research in that the estimate of α is relatively small (α = −0.07), albeit slightly negative (Charness & Rabin, 2002).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Charness & Rabin, 2002
Not checked in the free run.
- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
In the main study, the proportion of participants who shared with a human (29.93%) was significantly higher than participants who shared with machines.
“It reveals that the proportion of participants who shared with a human (29.93%) was significantly higher than participants who shared with machines.”
Quote in the paperSubstantial claim - Findingown data
All pairwise comparisons of the share of Option X between the fellow-human treatment and all other treatments are significant.
“All pairwise comparisons of the share of Option X using a Tukey’s range test between the fellow-human treatment and all other treatments are significant.”
Quote in the paperSubstantial claim - Findingown data
Cohen's d effect sizes for the comparison of fellow-human with token-player-behind-machine and programmer-behind-machine treatments were 0.070 and 0.097 in absolute value.
“We estimated Cohen’s ds of 0.070 and 0.097 in absolute value for the comparison with token-player-behind-machine treatment (24.31%) and programmer-behind-machine treatment (23.51%), respectively”
Quote in the paperNot a substantial claim - Findingown data
Compared with human-human settings, playing with a machine with no human beneficiary reduced the proportion of nonselfish choices by two thirds.
“Compared with human–human settings, playing with a machine with no human beneficiary reduced the proportion of nonselfish choices by two thirds”
Quote in the paperSubstantial claim - Findingown data
132 studies omitted information about the payoffs earned by machines.
“Fifth, 132 studies omitted the information about the payoffs earned by the machines altogether (e.g., Crandall et al., 2018; Ishowo-Oloko et al., 2019).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Crandall et al., 2018
Not checked in the free run.
- cites Ishowo-Oloko et al., 2019
Not checked in the free run.
- Findingown data
When cooperating with machines and no human earns the payoffs or when no information is provided, β is significantly smaller and effectively drops to 0 (-0.01 ≤ β ≤ 0.05; pairwise ps < .001, Cohen ds > 0.77 in absolute value).
“However, when cooperating with machines and no human earns the payoffs or when no information is provided, β is significantly smaller and effectively drops to 0 (–0.01 ≤ β ≤ 0.05, see also Fig. 4b; pairwise ps < .001, Cohen ds > 0.77 in absolute value).”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
People also showed less negative reciprocity (Cohen's ds > 0.50 in absolute value when comparing one of the three treatments involving a human beneficiary with one of the other three treatments; ps < .001 for all comparisons except for programmer vs. no information).
“People also showed less negative reciprocity (Cohen’s ds > 0.50 in absolute value when comparing one of the three treatments involving a human beneficiary with one of the other three treatments; ps < .001 for all comparisons except for programmer vs. no information).”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
In contrast to previous studies, the authors found a positive δ of 0.10.
“In contrast to previous studies, however, we found a positive d of 0.10.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
In these treatments, reciprocity disappeared altogether (-0.06 ≤ γ ≤ 0.007, -0.06 ≤ δ ≤ -0.03).
“In fact, in these treatments, reciprocity disappeared altogether (–0.06 ≤ g ≤ 0.007, –0.06 ≤ d ≤ –0.03; see Fig. 5).”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
When matched with machine partners, people reveal substantially stronger social preferences and reciprocity when they know that a human beneficiary receives the machine payoffs than when they know that no such "human behind the machine" exists.
“When matched with machine partners, people reveal substantially stronger social preferences and reciprocity when they know that a human beneficiary receives the machine payoffs than when they know that no such “human behind the machine” exists.”
Quote in the paperSubstantial claim - Findingown data
People show aversion to being ahead when playing with humans behind machines (token player: β = 0.24; programmer treatment: β = 0.25).
“Akin to human–human cooperation, people also show aversion to being ahead when playing with humans behind machines (token player: β = 0.24; programmer treatment: β = 0.25).”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
When comparing reciprocity preferences toward humans and machines, there were no differences in settings in which humans eventually earn the payoffs of the machine, for neither positive reciprocity nor negative reciprocity.
“When comparing reciprocity preferences toward humans and machines, again, we found no differences in settings in which humans eventually earn the payoffs of the machine, for neither positive reciprocity nor negative reciprocity.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
The estimate for α remains stable around 0 (-0.12 ≤ α ≤ 0.04) when playing with machines.
“When comparing a and β across treatments, we found that the estimate for a remains stable around 0 (–0.12 ≤ a ≤ 0.04) when playing with machines.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
In the fellow-human treatment, people generally show positive reciprocity (γ = 0.13).
“In the fellow-human treatment, our results replicate previous findings that people generally show positive reciprocity (g = 0.13).”
Quote in the paperNot a substantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
When participants are not informed about machine payoffs, the authors found weak social preferences toward machines.
“When participants are not informed about machine payoffs, we found weak social preferences toward machines.”
Quote in the paperSubstantial claim - Findingcitation
The estimate for β is much larger than the α estimate (β = 0.35).
“In line with previous work (Bruhin et al., 2018; Charness & Rabin, 2002; Klockmann et al., 2022a, 2022b), our estimate for β is much larger than the α estimate (β = 0.35).”
Quote in the paperNot a substantial claim0 of 3 sources checked- cites Bruhin et al., 2018
Not checked in the free run.
- cites Charness & Rabin, 2002
Not checked in the free run.
- cites Klockmann et al., 2022a, 2022b
Not checked in the free run.
- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Findingown data
The spiteful type showed no significant signs of positive or negative reciprocity.
“In addition, this group showed no significant signs of positive or negative reciprocity.”
Quote in the paperNot a substantial claim - Findingown data
The proportion of spiteful individuals is 10.74% when interacting with fellow humans.
“the proportion of spiteful individuals increases from 10.74% (fellow humans) to 15.33% when a token player earns the payoffs and to 20.67% when the algorithm programmer earns the payoffs”
Quote in the paperSubstantial claim - Findingown data
When interacting with a token player behind the machine, 58.00% are classified as the aheadness-averse/positive reciprocal type.
“programmer behind machine = 52.00%; token player behind machine = 58.00%”
Quote in the paperSubstantial claim - Findingown data
The mildly altruistic/selfish type is characterized by sharing relatively little in the dictator game.
“It is characterized by sharing relatively little in the dictator game.”
Quote in the paperNot a substantial claim - Findingown data
Only 3.81% of participants in the main study believed the payoffs would remain with the experimenter.
“Only very few participants (3.81%) believed the payoffs would remain with the experimenter.”
Quote in the paperNot a substantial claim - Findingown data
The spiteful type becomes the largest proportion in settings in which people cooperate with machines without a human beneficiary.
“Note that the spiteful type becomes the largest proportion of settings in which people cooperate with machines without a human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
The pattern of behavioral types not receiving any information mimics that of participants being told that the machine earns the payoffs most closely.
“Finally, the pattern of behavioral types not receiving any information mimics that of participants being told that the machine earns the payoffs most closely.”
Quote in the paperNot a substantial claim - Findingown data
The mildly altruistic/selfish type showed no signs of either positive or negative reciprocity.
“This group showed no signs of either positive or negative reciprocity.”
Quote in the paperNot a substantial claim - Findingown data
When the machine earns the profits, the share of spiteful individuals is 41.33%.
“When the machine earns the profits, the share of spiteful individuals increases to 41.33%”
Quote in the paperSubstantial claim - Findingown data
The third behavioral type describes 40.91% of participants.
“This type describes 40.91% of the participants.”
Quote in the paperNot a substantial claim - Findingown data
A "mildly altruistic/selfish" type comprises 28.10% of all participants.
“First is a “mildly altruistic/selfish” type that comprises 28.10% of all participants.”
Quote in the paperNot a substantial claim - Findingown data
When interacting with a programmer behind the machine, 52.00% are classified as the aheadness-averse/positive reciprocal type.
“programmer behind machine = 52.00%; token player behind machine = 58.00%”
Quote in the paperSubstantial claim - Findingown data
When it is explicitly stated that nobody receives the machine's payoffs, the share of spiteful individuals is 55.33%.
“when it is explicitly stated that nobody receives the machine’s payoffs, it even amounts to 55.33%”
Quote in the paperSubstantial claim - Findingown data
A model with three main behavioral types best fits the data.
“The analysis revealed that a model with three main behavioral types best fits our data”
Quote in the paperSubstantial claim- The authors state this as an assumption: The population comprises K distinct preference types, each characterized by its own set of parameters θ(k).
- Findingown data
A "spiteful" type accounts for 30.99% of the sample.
“The second behavioral type can be described as “spiteful” and accounts for 30.99% of the sample.”
Quote in the paperNot a substantial claim - Findingown data
The proportion of spiteful individuals is 20.67% when the algorithm programmer earns the payoffs.
“the proportion of spiteful individuals increases from 10.74% (fellow humans) to 15.33% when a token player earns the payoffs and to 20.67% when the algorithm programmer earns the payoffs”
Quote in the paperSubstantial claim - Findingown data
The spiteful type revealed strong negative aheadness and behindness aversion.
“It revealed strong negative aheadness and behindness aversion”
Quote in the paperNot a substantial claim - Findingown data
When interacting with humans, 73.15% of participants are classified as the aheadness-averse/positive reciprocal-behavioral type.
“When interacting with humans, almost three quarters of participants (73.15%) are classified as the aheadness-averse/positive reciprocal-behavioral type.”
Quote in the paperSubstantial claim - Findingown data
The proportion of spiteful individuals is 15.33% when a token player earns the payoffs.
“the proportion of spiteful individuals increases from 10.74% (fellow humans) to 15.33% when a token player earns the payoffs and to 20.67% when the algorithm programmer earns the payoffs”
Quote in the paperSubstantial claim - Findingown data
The mildly altruistic/selfish type showed positive aheadness and behindness aversion.
“This group also showed positive aheadness and behindness aversion.”
Quote in the paperNot a substantial claim - Findingown data
The distribution of these types differed starkly across treatments.
“The distribution of these types differed starkly across treatments.”
Quote in the paperSubstantial claim - Findingown data
The change in decisions was observed in all three treatments, even if a human eventually earns the machines' payoffs.
“We observed this change in decisions in all three treatments, even if a human eventually earns the machines’ payoffs.”
Quote in the paperSubstantial claim - Findingown data
For allocation decisions, participants showed the highest levels of sharing propensity when sharing with fellow humans.
“For allocation decisions, participants showed the highest levels of sharing propensity when sharing with fellow humans.”
Quote in the paperSubstantial claim - Findingown data
The reciprocity norm did not extend to machine partners without a human beneficiary.
“However, this reciprocity norm did not extend to machine partners without a human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
The belief that the money stayed with the experimenter was 21.42% in the follow-up study compared with 3.81% in the main study.
“the belief that the money stayed with the experimenter was much more frequent in the follow-up study (21.42%) compared with the main study (3.81%).”
Quote in the paperNot a substantial claim - Findingown data
The degree to which people differentiated between humans and machines varied according to the specific social-preference parameter.
“Analyzing the full set of social-preference parameters revealed that the degree to which people differentiated between humans and machines varied according to the specific parameter.”
Quote in the paperSubstantial claim - Findingown data
Choosing the nonselfish option significantly dropped when matched with a machine receiver that had a human beneficiary behind it.
“Choosing the nonselfish option significantly dropped when matched with a machine receiver that had a human beneficiary behind it.”
Quote in the paperSubstantial claim - Findingown data
People did not seem to mind an advantageous inequality in their favor that comes at a cost for a machine without a human beneficiary.
“Yet people did not seem to mind an advantageous inequality in their favor that comes at a cost for a machine without a human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
Among uninvolved participants in the follow-up study, 18.18% believed the money was not paid out at all.
“Among uninvolved participants in the follow-up study, a much smaller fraction believed the money was not paid out at all (18.18%).”
Quote in the paperNot a substantial claim - Findingown data
The proportion of people who believe the money was not paid out differs significantly between the main study and the follow-up study, χ2(1) = 16.34, p < .001.
“A χ2 test with Yates’s continuity correction comparing the proportion of people who believe the money was not paid out in the main study and the follow-up study, respectively, revealed a significant difference, χ2(1) = 16.34, p < .001.”
Quote in the paperNot a substantial claim - Findingown data
When interacting with machines and no human earned the payoffs, most people acted spitefully.
“Yet when interacting with machines and no human earned the payoffs, most people acted spitefully.”
Quote in the paperSubstantial claim - Findingown data
Altruistic giving was further reduced for machine counterparts without any human beneficiary.
“Altruistic giving was further reduced for machine counterparts without any human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
For positive (γ) and negative (δ) reciprocity, people were sensitive to paying back humans and machines with a human beneficiary.
“Estimates for positive (γ) and negative (δ) reciprocity revealed a similar pattern: People were sensitive to paying back humans and machines with a human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
For behindness aversion (α), there were no differences whatsoever.
“For behindness aversion (α), we found no differences whatsoever.”
Quote in the paperSubstantial claim - Findingown data
The finite mixture-estimation model identified three types of preferences.
“Our finite mixture-estimation model identified three types of preferences.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The population comprises K distinct preference types, each characterized by its own set of parameters θ(k).
- Findingown data
The difference across studies in belief that money stayed with the experimenter is significant, χ2(1) = 14.42, p < .001.
“This difference across studies is significant as well, χ2(1) = 14.42, p < .001.”
Quote in the paperNot a substantial claim - Findingown data
A human beneficiary "behind the machine" can increase giving rates compared with settings without a human beneficiary.
“we showed that a human beneficiary “behind the machine” can increase giving rates compared with settings without a human beneficiary.”
Quote in the paperSubstantial claim - Findingown data
For aheadness aversion (β), people were averse to being ahead of fellow humans and machines when humans earn the payoffs.
“for aheadness aversion (β), we found that people were averse to being ahead of fellow humans and machines when humans earn the payoffs.”
Quote in the paperSubstantial claim - Findingown data
In our study, this tendency was observed for a large fraction of participants only when a human behind the machine earned the payoffs.
“In our study, this tendency was observed for a large fraction of participants only when a human behind the machine earned the payoffs.”
Quote in the paperSubstantial claim - Findingown data
People also adjust their behavior according to different implementations of machine partners.
“Here, we found evidence that people also adjust their behavior according to different implementations of machine partners.”
Quote in the paperSubstantial claim - Findingown data
In our study, social preferences mostly vanished in the treatments in which the payoffs were not paid out at all (no one earns) or paid to the machine (machine earns), which both resulted in the partner's payoffs being effectively zero.
“In our study, social preferences mostly vanished in the treatments in which the payoffs were not paid out at all (no one earns) or paid to the machine (machine earns), which both resulted in the partner’s payoffs being effectively zero.”
Quote in the paperSubstantial claim - Findingown data
In the follow-up study, the proportion believing the money was not paid out dropped significantly.
“In the follow-up study, in which people had no motivation to adhere to this particular belief, the proportion dropped significantly.”
Quote in the paperSubstantial claim - Findingown data
In the no-information settings, people acted as selfishly as they did in the treatments in which the money was paid to a machine or explicitly not paid out at all.
“In the no-information settings, people acted as selfishly as they did in the treatments in which the money was paid to a machine or explicitly not paid out at all.”
Quote in the paperSubstantial claim - Findingown data
In these treatments, only a small fraction of people forwent their own payoffs to benefit their partner or to destroy their payoff.
“In these treatments, we found that only a small fraction of people forwent their own payoffs to benefit their partner or to destroy their payoff.”
Quote in the paperSubstantial claim - Findingown data
Answers in the open-text field in the main study suggest that most people believe the money is not paid out.
“Answers in the open-text field in our main study regarding beliefs about the machine payoffs suggest that most people, in fact, believe the money is not paid out.”
Quote in the paperSubstantial claim - Findingown data
Our findings show that the no-information implementation consistently leads to weak revealed social preferences toward machines.
“Our findings show that this implementation consistently leads to weak revealed social preferences toward machines.”
Quote in the paperSubstantial claim - Findingown data
People's preferences toward machines differ depending on the machines' payoffs and the information provided about them.
“Here, we present a systematic behavioral investigation revealing that people’s preferences toward machines differ depending on the machines’ payoffs and the information provided about them.”
Quote in the paperSubstantial claim - Findingown data
People reveal systematically higher social preferences toward machines when they know that a human beneficiary receives the payoffs accrued by the machine compared with when they know that no such human behind the machine exists.
“People reveal systematically higher social preferences toward machines when they know that a human beneficiary receives the payoffs accrued by the machine compared with when they know that no such human behind the machine exists.”
Quote in the paperSubstantial claim - Findingown data
For none of the social-preference indicators, there was no significant difference between the setting in which an uninvolved token player or the algorithm programmer earned the payoffs.
“For none of the social-preference indicators, we found no significant difference between the setting in which an uninvolved token player or the algorithm programmer earned the payoffs.”
Quote in the paperSubstantial claim - Premisecitation
A comparative analysis of how different machine-payoff implementations influence social preferences is lacking.
“A comparative analysis of how these different implementations of machine players in economic games influence social preferences is lacking, undermining the understanding of how cooperation with machines differs from cooperation among humans (Dafoe et al., 2020).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Dafoe et al., 2020
Not checked in the free run.
- Interpretationown data
The observed behavioral patterns indicate that participants still had efficiency concerns toward a machine even if no human beneficiary existed.
“The observed behavioral patterns indicate that participants still had efficiency concerns toward a machine even if no human beneficiary existed.”
Quote in the paperSubstantial claim - Interpretationown data
The results of the main study indicate that people show signs of reciprocity toward fellow humans and machines in settings where humans earn the payoffs.
“In conclusion, the results of the main study indicate that people show signs of reciprocity toward fellow humans and machines in settings where humans earn the payoffs.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationreasoning
For human partners, people appear to be particularly forgiving when faced with an unkind action by the first mover.
“For human partners, this means that people appear to be particularly forgiving when faced with an unkind action by the first mover.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationown data
The extent to which humans cooperate with machines depends on the implementation and information about the machine's earnings.
“Thus, our results suggest that the extent to which humans cooperate with machines depends on the implementation and information about the machine’s earnings.”
Quote in the paperSubstantial claim - Interpretationown data
People are generally behindness averse independent of whether they interact with humans or machines.
“The results for the a and β estimates indicate that people are generally behindness averse independent of whether they interact with humans or machines.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationcitation
The general envy toward machines does not depend on the realization of the payoffs.
“These findings extend previous work by showing that the general envy toward machines does not depend on the realization of the payoffs (De Melo et al., 2016).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites De Melo et al., 2016
Not checked in the free run.
- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationown data
People form their beliefs about machine payoffs in a self-serving way.
“Comparing survey answers with those from a follow-up study (N = 150), we conclude that people form their beliefs about machine payoffs in a self-serving way.”
Quote in the paperSubstantial claim - Interpretationown data
People show aheadness aversion only when cooperating with humans or humans behind machines, not when matched with machines without a human beneficiary.
“In line with the literature showing that people do not feel guilty about exploiting machines, we also discover that people show aheadness aversion only when cooperating with humans or humans behind machines, not when matched with machines without a human beneficiary.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationown data
In disadvantageous settings, positive other-regarding preferences disappear when cooperating with machines void of a human beneficiary.
“Hence, in disadvantageous settings, positive other-regarding preferences disappear when cooperating with machines void of a human beneficiary.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationown data
The authors add nuance to the previously observed general finding that people reciprocate machines by highlighting that such reciprocity depends on the realization of the machine payoffs.
“Hence, we add nuance to the previously observed general finding that people reciprocate machines (Pratt et al., 2007; Sandoval et al., 2016) by highlighting that such reciprocity depends on the realization of the machine payoffs.”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Pratt et al., 2007
Not checked in the free run.
- cites Sandoval et al., 2016
Not checked in the free run.
- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
- Interpretationown data
The mere inclusion of a machine as an intermediary reduced people's willingness to share their endowment.
“The mere inclusion of a machine as an intermediary reduced people’s willingness to share their endowment.”
Quote in the paperSubstantial claim - Interpretationown data
How revealed social preferences toward machines differ from interhuman social preferences depends strongly on the machine-payoff implementation and the type of social preferences.
“How revealed social preferences toward machines differ from interhuman social preferences depends strongly on the machine-payoff implementation and the type of social preferences.”
Quote in the paperSubstantial claim - Interpretationown data
People generally dislike being behind, independent of their partner.
“This shows that people generally dislike being behind, independent of their partner.”
Quote in the paperSubstantial claim - Interpretationreasoning
Because providing no information is the most common implementation in behavioral science, studies without payoff information might lead to weak social-preference estimations.
“Because providing no information is the most common implementation in behavioral science, studies without payoff information might lead to weak social-preference estimations.”
Quote in the paperSubstantial claim - Premisecitation
A plausible explanation receiving initial empirical support is that people often act toward machines as if they were humans.
“A plausible explanation receiving initial empirical support reads that people often act toward machines as if they were humans (Nass et al., 1994; Nielsen et al., 2022).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Nass et al., 1994
Not checked in the free run.
- cites Nielsen et al., 2022
Not checked in the free run.
- Premisecitation
A machine player by no means always eliminates social preferences as they were originally intended to.
“A machine player by no means always eliminates social preferences as they were originally intended to (Andreoni, 1995; Houser & Kurzban, 2002).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Andreoni, 1995
Not checked in the free run.
- cites Houser & Kurzban, 2002
Not checked in the free run.
- Interpretationcitation
People change their behavioral profile contingent on their counterparts.
“It adds credence to the notion that people change their behavioral profile contingent on their counterparts, as research on conditional cooperation has shown (Burton-Chellew et al., 2016; Fischbacher et al., 2001).”
Quote in the paperSubstantial claim0 of 2 sources checked- cites Burton-Chellew et al., 2016
Not checked in the free run.
- cites Fischbacher et al., 2001
Not checked in the free run.
- Interpretationown data
The way in which machine players are incorporated strongly influences how people reveal social preferences toward them.
“Second, the way in which these machine players are incorporated strongly influences how people reveal social preferences toward them.”
Quote in the paperSubstantial claim - Interpretationown data
The information gap appears to allow people to form motivated beliefs about the payoffs.
“Hence, our results indicate that the information gap appears to allow people to form motivated beliefs about the payoffs.”
Quote in the paperSubstantial claim - Interpretationown data
Whether a deserving or undeserving human earns these payoffs seems less relevant.
“Whether a deserving or undeserving human earns these payoffs seems less relevant.”
Quote in the paperSubstantial claim - Interpretationown data
Lack of information and thus uncertainty about an eventual human beneficiary might lead people to cooperate less with machines compared with when the actual (human) beneficiary is disclosed.
“Lack of information and thus uncertainty about an eventual human beneficiary might lead people to cooperate less with machines compared with when the actual (human) beneficiary is disclosed.”
Quote in the paperSubstantial claim
- May combine two claims. 16
- Classified as probably not a substantial claim. 27
May combine two claims.
Findingown dataThe authors categorized the instructions of 160 behavioral studies reviewed by March (2021) and found five main ways of machine-payoff implementations.
“We systematically categorized the instructions used in 160 behavioral studies reviewed by March (2021) and found five main ways of such machine-payoff implementations.”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites March (2021)
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
Findingown dataIn five experiments, participants were informed that the machine's earnings would be burnt.
“Fourth and relatedly, the experimental instructions of five experiments informed participants that the machine’s earnings would be “burnt.””
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
FindingcitationTwo studies passed machine payoffs to the programmer of the algorithm.
“This implementation has been used in (only) two studies in which the machine players’ payoffs were passed on to the programmer of the algorithm (Azaria et al., 2016; Lin et al., 2014).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Azaria et al., 2016
Not checked in the free run.
- cites Lin et al., 2014
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
FindingcitationSeven studies used a human beneficiary not involved in the task (token player) as receiver of machine payoffs.
“Second, another group of seven studies also used a human beneficiary who, however, was not involved in the task. These studies use a so-called token player as the eventual receiver of the machine payoffs (Köbis & Mossink, 2021).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Köbis & Mossink, 2021
Not checked in the free run.
May combine two claims.
Findingown dataThe likelihood of choosing the nonselfish option was highest toward fellow humans, followed by humans behind machines, and lowest toward machines without any human beneficiary.
“Overall, we found a three-stepped pattern of results: The likelihood of choosing the nonselfish option was highest toward fellow humans, followed by humans behind machines, and lowest toward machines without any human beneficiary.”
Quote in the paperSubstantial claimMay combine two claims.
Findingown dataWhen people cooperate with machines and no human earns the payoffs or no information is provided about the payoffs, people showed significantly less positive reciprocity (ps < .001 and Cohen's ds > 0.71 in absolute value when comparing the fellow-human treatment with all three treatments).
“People showed significantly less positive reciprocity (ps < .001 and Cohen’s ds > 0.71 in absolute value when comparing the fellow-human treatment with all three treatments).”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
May combine two claims.
Findingown dataThe percentage of mildly altruistic/selfish types fluctuates between 16.11% and 36.00% and remains the second-largest group across all treatments.
“The percentage of mildly altruistic/selfish types, although fluctuating between 16.11% and 36.00%, remains the second-largest group across all treatments.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataThe third behavioral type is aheadness averse and positively reciprocal.
“The third type is “aheadness averse and positively reciprocal,” hence people who dislike unfair outcomes that favor themselves and answer kind acts with kindness.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataIn the main study, the modal answer coding in the no-information treatment is "No one; the money is not paid out," with 41.9% of responses indicating the belief.
“the modal answer coding in the main study is “No one; the money is not paid out,” with 41.9% of responses indicating the belief.”
Quote in the paperNot a substantial claimMay combine two claims.
Findingown dataWhen interacting with machines, most people disliked favorable inequality and showed positive reciprocity, reflecting typical behavior when interacting with (unknown) others.
“When interacting with machines, most people disliked favorable inequality and showed positive reciprocity, reflecting typical behavior when interacting with (unknown) others.”
Quote in the paperSubstantial claimMay combine two claims.
FindingreasoningThe estimated parameters of aheadness and behindness aversion, reciprocity, and fair-mindedness are all close to zero on average in these treatments.
“In line with this view, the estimated parameters of aheadness and behindness aversion, reciprocity, and fair-mindedness are all close to zero on average.”
Quote in the paperSubstantial claimMay combine two claims.
FindingreasoningNot informing people about the machine payoffs leads to low social preferences as people form beliefs about the payoffs in a self-favoring way.
“Not informing people about the machine payoffs leads to low social preferences as people form beliefs about the payoffs in a self-favoring way.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationMore than 160 studies from various disciplines have reported on how people cooperate with machines in behavioral experiments.
“By now, more than 160 studies from various disciplines have reported on how people cooperate with machines in behavioral experiments.”
- Also stated in: § Different Implementations of Machine Players · plain
“We systematically categorized the instructions used in 160 behavioral studies reviewed by March (2021)”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites March (2021)
Not checked in the free run.
- Also stated in: § Different Implementations of Machine Players · plain
Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.
InterpretationcitationThis behavior most likely stemmed from confusion or reputation concerns toward the experimenter.
“This behavior most likely stemmed from confusion (Burton-Chellew et al., 2016; Nielsen et al., 2021) or reputation concerns toward the experimenter.”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Burton-Chellew et al., 2016
Not checked in the free run.
- cites Nielsen et al., 2021
Not checked in the free run.
May combine two claims.
Interpretationown dataReciprocity disappears when there is no human beneficiary or a void of information about the payoffs.
“Yet reciprocity disappears when there is no human beneficiary or a void of information about the payoffs.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
May combine two claims.
InterpretationreasoningThe small positive δ (especially in light of a larger, positive γ) indicates a general positive weight on the other player's payoffs and not necessarily a rewarding or direct positive reaction to an unkind move.
“Therefore, the small positive d (especially in light of a larger, positive g) indicates a general positive weight on the other player’s payoffs and not necessarily a rewarding or direct positive reaction to an unkind move.”
Quote in the paperSubstantial claim- The authors state this as an assumption: The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
- The authors state this as an assumption: The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
Classified as probably not a substantial claim.
Study design and conductown dataParticipants were randomly assigned to one of six between-subjects treatments.
“We randomly assigned participants to one of six between-subjects treatments.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductnone givenThe study was conducted in a fully incentivized way.
“We conducted the study in a fully incentivized way.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
Study design and conductnone givenOne of the 30 rounds was randomly chosen and paid out.
“Participants learned that one of the 30 rounds would be randomly chosen and paid out.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Study design and conductnone givenAll materials, data code to reproduce all analyses, and figures are openly available.
“All materials, data code to reproduce all analyses, and figures are openly available at https://osf.io/efmvh/?view_only=fbba5877d7014052adbbf404703eb1ff.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Study design and conductreasoningEach participant played one type of game in one treatment to avoid carryover effects between the games.
“Each participant played one type of game in one treatment to avoid carryover effects between the games.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.
Study design and conductnone givenAll data were collected under ethics approval and after obtaining informed consent.
“All data were collected under ethics approval and after obtaining informed consent.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenA research assistant naive to the study's purpose coded the answers in the no-information treatment according to five categories.
“A research assistant naive to the study’s purpose coded the answers in the no-information treatment according to five categories.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Study design and conductnone givenA new research assistant naive to the study's purpose coded the answers using the same coding scheme as in the main study.
“We asked a new research assistant naive to the study’s purpose to code the answers using the same coding scheme as in the main study.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationPrevious work has shown the basic pattern of lower altruistic preferences for machines versus humans.
“Although previous work has shown the basic pattern of lower altruistic preferences for machines versus humans (Burton-Chellew et al., 2016; Nielsen et al., 2021)”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Burton-Chellew et al., 2016
Not checked in the free run.
- cites Nielsen et al., 2021
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationPrevious research showed such intermediaries can lead to different, often less prosocial behaviors.
“This matches previous research that showed such intermediaries can lead to different, often less prosocial behaviors (Hsee et al., 2003; Mazar et al., 2008).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Hsee et al., 2003
Not checked in the free run.
- cites Mazar et al., 2008
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationPrevious research has shown that people even reveal social preferences toward machines when they are fully aware of the game's logic.
“Previous research, however, has shown that people even reveal social preferences toward machines when they are fully aware of the game’s logic (Nielsen et al., 2021).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Nielsen et al., 2021
Not checked in the free run.
Probably not a substantial claim: may not pass S4 (based).
Premisenone givenSo far, researchers have diverged substantially in how they implement (the payoffs of) machine players in economic games.
“So far, researchers have diverged substantially in how they implement (the payoffs of) machine players in economic games.”
Quote in the paperNot a substantial claimMay combine two claims.
InterpretationcitationPeople in the main study (vs. the follow-up study) had an incentive to believe that no human behind the machine earned the payoffs because this setting enabled them to justify acting more selfishly, possibly because it dampened the guilt of selfishness.
“In line with belief-based utility (Loewenstein & Molnar, 2018), people in the main study (vs. the follow-up study) had an incentive to believe that no human behind the machine earned the payoffs because this setting enabled them to justify acting more selfishly, possibly because it dampened the guilt of selfishness.”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Loewenstein & Molnar, 2018
Not checked in the free run.
Classified as probably not a substantial claim.
Premisenone givenThe approach most commonly used has been not to provide any information about the machine's payoffs.
“So far, the approach most commonly used has been not to provide any information about the machine’s payoffs.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationPrevious research on human-human cooperation showed prosocial behavior drops when uncertainty about the outcome of one's actions exists.
“Our findings align with previous research on human–human cooperation that showed prosocial behavior drops when uncertainty about the outcome of one’s actions exists (Kappes et al., 2018).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Kappes et al., 2018
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
InterpretationcitationAnother plausible explanation is that participants perceive uninvolved participants who, like them, are recruited from an online click-work platform like Prolific as equally deserving as the programmer.
“Another plausible explanation is that participants perceive uninvolved participants who, like them, are recruited from an online click-work platform like Prolific as equally deserving as the programmer.”
Quote in the paperNot a substantial claimClassified as probably not a substantial claim.
Premisenone givenParticipants received no information about the machines' payoffs in these experiments.
“Note that participants received no information about the machines’ payoffs in these experiments.”
Quote in the paperSubstantial claimClassified as probably not a substantial claim.
PremisecitationWhen people know they interact with a machine (vs. a human), they cooperate less.
“When people know they interact with a machine (vs. a human), they cooperate less.”
Quote in the paperSubstantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
Interpretationown dataThe machines acting as intermediaries might have mooted the well-established deservingness heuristic.
“The machines acting as intermediaries might have mooted the well-established deservingness heuristic (Petersen et al., 2011).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Petersen et al., 2011
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).
InterpretationreasoningTransparency about a human beneficiary of the machines' earnings might help to resolve the transparency-efficiency paradox.
“Paradoxically, transparency (about a human beneficiary of the machines’ earnings) might help to resolve the transparency–efficiency paradox.”
Quote in the paperNot a substantial claimProbably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationEmpirical research showed online participants treat other participants from the same online platform preferentially compared with random strangers.
“This notion receives support from empirical research that showed online participants treat other participants from the same online platform preferentially (compared with random strangers; Almaatouq et al., 2020).”
Quote in the paperNot a substantial claim0 of 1 sources checked- cites Almaatouq et al., 2020
Not checked in the free run.
Probably not a substantial claim: may not pass S3 (load-bearing).
PremisecitationTheoretical work has argued machines increasingly acting as intermediaries and delegates might reduce human social and moral concerns.
“This explanation aligns with theoretical work that has argued machines increasingly acting as intermediaries and delegates might reduce human social and moral concerns (Gratch & Fast, 2022; Köbis et al., 2021).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Gratch & Fast, 2022
Not checked in the free run.
- cites Köbis et al., 2021
Not checked in the free run.
Classified as probably not a substantial claim.
PremisecitationPrevious research on human and machine partners has uncovered a transparency-efficiency paradox.
“Previous research on human and machine partners has uncovered a transparency–efficiency paradox (Ishowo-Oloko et al., 2019).”
Quote in the paperSubstantial claim0 of 1 sources checked- cites Ishowo-Oloko et al., 2019
Not checked in the free run.
May combine two claims.
PremisecitationAI systems imbued with more lasting memory capacities are beginning to appear, enabling repeated interactions with AI systems.
“However, AI systems imbued with more lasting memory capacities are beginning to appear (Saravanan et al., 2022), enabling repeated interactions with AI systems, which is a key step toward turning them into companions (Krämer et al., 2011).”
Quote in the paperNot a substantial claim0 of 2 sources checked- cites Saravanan et al., 2022
Not checked in the free run.
- cites Krämer et al., 2011
Not checked in the free run.
The authors state this as an assumption.
The random-utility model assumes that utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.
“This method assumes that the utility consists of a deterministic utility (behavioral model) and a random component representing noise in the utility evaluation.”
The authors state this as an assumption.
The assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.
“By exploiting the binary decision structure in both games, the assumption is that a participant chooses Option X if the random utility exceeds the utility of Option Y to estimate revealed social-preference parameters.”
The authors state this as an assumption.
The population comprises K distinct preference types, each characterized by its own set of parameters θ(k).
“Here, we assumed that the population comprises K distinct preference types, each characterized by its own set of parameters θ(k).”
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?
Prosocial behavior depends on people's beliefs about the relationship between the machine and the humans behind it.
“Prosocial behavior depends on people's beliefs about the relationship between the machine and the humans behind it (51).”
From The impact of generative artificial intelligence on socioeconomic inequalities and policy making (2024) · cites this paper as [51]
Statistics recomputed
Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.
- Consistentχ²(1) = 16.34reported p < .001 · recomputed p < .001
“χ2(1) = 16.34, p < .001”
- Consistentχ²(1) = 14.42reported p < .001 · recomputed p < .001
“χ2(1) = 14.42, p < .001”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Discussion reverses the finding stated in a cited reference's title
MinorThe discussion says uncertainty makes prosocial behavior drop, while the paper's reference title says uncertainty increases prosocial behaviour.
“Our findings align with previous research on human–human cooperation that showed prosocial behavior drops when uncertainty about the outcome of one’s actions exists (Kappes et al., 2018).”
“Kappes A. Nussberger A.-M. Faber N. S. Kahane G. Savulescu J. Crockett M. J. (2018). Uncertainty about the impact of social decisions increases prosocial behaviour.”
Cites
Axelrod, R., & Hamilton, W. D. (1981) · Science · doi:10.1126/science.7466396 · Cited by 20,065 · see citing works
Advances in Neural Information Processing Systems 29
Teymur, O., Zygalakis, K., & Calderhead, B. (2016) · Cited by 13,881 · see citing works
Duflo, E. (2019) · AEA Papers and Proceedings · doi:10.1257/pandp.109.612 · Cited by 11,235 · see citing works
Mastering the game of Go without human knowledge
Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., Hubert, T., Baker, L., Lai, M., Bolton, A., et al. (2017) · Nature · doi:10.1038/nature24270 · Cited by 9,084 · see citing works
The Quarterly Journal of Economics
Kuznets, S. (2024) · Routledge eBooks · doi:10.4324/9781003549741-8 · Cited by 7,177 · see citing works
Dissecting racial bias in an algorithm used to manage the health of populations
Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019) · Science · doi:10.1126/science.aax2342 · Cited by 7,137 · see citing works
Cooperation and Punishment in Public Goods Experiments
Fehr, E., & Gächter, S. (2000) · American Economic Review · doi:10.1257/aer.90.4.980 · Cited by 4,416 · see citing works
Machines and Mindlessness: Social Responses to Computers
Nass, C., & Moon, Y. (2000) · Journal of Social Issues · doi:10.1111/0022-4537.00153 · Cited by 4,022 · see citing works
Understanding Social Preferences with Simple Tests
Charness, G., & Rabin, M. (2002) · The Quarterly Journal of Economics · doi:10.1162/003355302760193904 · Cited by 3,970 · see citing works
A few useful things to know about machine learning
Domingos, P. (2012) · Communications of the ACM · doi:10.1145/2347736.2347755 · Cited by 3,326 · see citing works
The Dishonesty of Honest People: A Theory of Self-Concept Maintenance
Mazar, N., Amir, O., & Ariely, D. (2008) · Journal of Marketing Research · doi:10.1509/jmkr.45.6.633 · Cited by 2,875 · see citing works
Are people conditionally cooperative? Evidence from a public goods experiment
Fischbacher, U., Gächter, S., & Fehr, E. (2001) · Economics Letters · doi:10.1016/s0165-1765(01)00394-9 · Cited by 2,780 · see citing works
Nass, C., Steuer, J., & Tauber, E. R. (1994) · Proceedings of the SIGCHI Conference on Human Factors in Computing Systems · doi:10.1145/191666.191703 · Cited by 2,148 · see citing works
Awad, E., Dsouza, S., Kim, R., Schulz, J., Henrich, J., Shariff, A., Bonnefon, J.-F., & Rahwan, I. (2018) · Nature · doi:10.1038/s41586-018-0637-6 · Cited by 1,859 · see citing works
The social dilemma of autonomous vehicles
Bonnefon, J.-F., Shariff, A., & Rahwan, I. (2016) · Science · doi:10.1126/science.aaf2654 · Cited by 1,527 · see citing works
Emergence of cooperation and evolutionary stability in finite populations
Nowak, M. A., Sasaki, A., Taylor, C., & Fudenberg, D. (2004) · Nature · doi:10.1038/nature02414 · Cited by 1,403 · see citing works
Rahwan, I., Cebrian, M., Obradovich, N., Bongard, J., Bonnefon, J.-F., Breazeal, C., Crandall, J. W., Christakis, N. A., Couzin, I. D., Jackson, M. O., et al. (2019) · Nature · doi:10.1038/s41586-019-1138-y · Cited by 1,364 · see citing works
Mastering Atari, Go, chess and shogi by planning with a learned model
Schrittwieser, J., Antonoglou, I., Hubert, T., Simonyan, K., Sifre, L., Schmitt, S., Guez, A., Lockhart, E., Hassabis, D., Graepel, T., et al. (2020) · RePEc: Research Papers in Economics · Cited by 1,310 · see citing works
Villani, V., Pini, F., Leali, F., & Secchi, C. (2018) · Mechatronics · doi:10.1016/j.mechatronics.2018.02.009 · Cited by 1,278 · see citing works
On the Morality of Artificial Agents
Floridi, L., & Sanders, J. W. (2004) · Minds and Machines · doi:10.1023/b:mind.0000035461.63578.9d · Cited by 1,218 · see citing works
Mastering Atari, Go, chess and shogi by planning with a learned model
Schrittwieser, J., Antonoglou, I., Hubert, T., Simonyan, K., Sifre, L., Schmitt, S., Guez, A., Lockhart, E., Hassabis, D., Graepel, T., et al. (2020) · Nature · doi:10.1038/s41586-020-03051-4 · Cited by 1,214 · see citing works
Human Decisions and Machine Predictions
Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., & Mullainathan, S. (2018) · The Quarterly Journal of Economics · doi:10.1093/qje/qjx032 · Cited by 1,040 · see citing works
What ChatGPT and generative AI mean for science
Stokel-Walker, C., & Van Noorden, R. (2023) · Nature · doi:10.1038/d41586-023-00340-6 · Cited by 828 · see citing works
Personality and prosocial behavior: A theoretical framework and meta-analysis.
Thielmann, I., Spadaro, G., & Balliet, D. P. (2019) · Psychological Bulletin · doi:10.1037/bul0000217 · Cited by 689 · see citing works
Building a Stronger CASA: Extending the Computers Are Social Actors Paradigm
Gambino, A., Fox, J., & Ratan, R. (2020) · Human-Machine Communication · doi:10.30658/hmc.1.5 · Cited by 626 · see citing works
Measuring and Bounding Experimenter Demand
de Quidt, J., Haushofer, J., & Roth, C. (2018) · American Economic Review · doi:10.1257/aer.20171330 · Cited by 573 · see citing works
Social behavior for autonomous vehicles
Schwarting, W., Pierson, A. N., Alonso–Mora, J., Karaman, S., & Rus, D. L. (2019) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1820676116 · Cited by 495 · see citing works
Nonzero: the logic of human destiny
(2000) · Choice Reviews Online · doi:10.5860/choice.37-6234 · Cited by 488 · see citing works
Detecting Failures of Backward Induction: Monitoring Information Search in Sequential Bargaining
Johnson, E. J., Camerer, C. F., Sen, S., & Rymon, T. (2002) · Journal of Economic Theory · doi:10.1006/jeth.2001.2850 · Cited by 411 · see citing works
Köbis, N., & Mossink, L. D. (2021) · Computers in Human Behavior · doi:10.1016/j.chb.2020.106553 · Cited by 406 · see citing works
Self-citationUnderstanding, explaining, and utilizing medical artificial intelligence
Cadario, R., Longoni, C., & Morewedge, C. K. (2021) · Nature Human Behaviour · doi:10.1038/s41562-021-01146-0 · Cited by 342 · see citing works
Psychological roadblocks to the adoption of self-driving vehicles
Shariff, A., Bonnefon, J.-F., & Rahwan, I. (2017) · Nature Human Behaviour · doi:10.1038/s41562-017-0202-6 · Cited by 336 · see citing works
Petersen, M. B., Slothuus, R., Stubager, R., & Togeby, L. (2010) · European Journal of Political Research · doi:10.1111/j.1475-6765.2010.01923.x · Cited by 311 · see citing works
Laitinen, H. A. (1970) · Analytical Chemistry · doi:10.1021/ac60288a601 · Cited by 292 · see citing works
Revisiting Kindness and Confusion in Public Goods Experiments
Houser, D. E., & Kurzban, R. (2002) · American Economic Review · doi:10.1257/00028280260344605 · Cited by 280 · see citing works
Crandall, J. W., Oudah, M., Tennom, Ishowo-Oloko, F., Abdallah, S., Bonnefon, J., Cebrián, M., Shariff, A., Goodrich, M. A., & Rahwan, I. (2018) · Nature Communications · doi:10.1038/s41467-017-02597-8 · Cited by 240 · see citing works
Cooperative AI: machines must learn to find common ground
Dafoe, A., Bachrach, Y., Hadfield, G., Horvitz, E., Larson, K., & Graepel, T. (2021) · Nature · doi:10.1038/d41586-021-01170-0 · Cited by 232 · see citing works
Bad machines corrupt good morals
Köbis, N., Bonnefon, J., & Rahwan, I. (2021) · Nature Human Behaviour · doi:10.1038/s41562-021-01128-2 · Cited by 209 · see citing works
Self-citationBehavioural evidence for a transparency–efficiency tradeoff in human–machine cooperation
Ishowo-Oloko, F., Bonnefon, J.-F., Soroye, Z., Crandall, J., Rahwan, I., & Rahwan, T. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0113-5 · Cited by 178 · see citing works
Hsee, C. K., Yu, F., Zhang, J., & Zhang, Y. (2003) · Journal of Consumer Research · doi:10.1086/374702 · Cited by 171 · see citing works
Sandoval, E. B., Brandstetter, J., Obaid, M., & Bartneck, C. (2015) · International Journal of Social Robotics · doi:10.1007/s12369-015-0323-x · Cited by 142 · see citing works
Conditional cooperation and confusion in public-goods experiments
Burton-Chellew, M. N., Mouden, C. E., & West, S. A. (2016) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1509740113 · Cited by 142 · see citing works
The distributional preferences of an elite
Fisman, R. J., Jakiela, P., Kariv, S., & Markovits, D. B. (2015) · Science · doi:10.1126/science.aab0096 · Cited by 138 · see citing works
How users reciprocate to computers
Fogg, B. J., & Nass, C. (1997) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/1120212.1120419 · Cited by 133 · see citing works
The promise and perils of using artificial intelligence to fight corruption
Köbis, N., Starke, C., & Rahwan, I. (2022) · Nature Machine Intelligence · doi:10.1038/s42256-022-00489-1 · Cited by 112 · see citing works
Self-citationThe many Faces of Human Sociality: Uncovering the Distribution and Stability of Social Preferences
Bruhin, A., Fehr, E., & Schunk, D. (2018) · Journal of the European Economic Association · doi:10.1093/jeea/jvy018 · Cited by 108 · see citing works
Oliveira, R., Arriaga, P., Santos, F. P., Mascarenhas, S. F., & Paiva, A. (2020) · Computers in Human Behavior · doi:10.1016/j.chb.2020.106547 · Cited by 100 · see citing works
Krämer, N. C., Eimler, S. C., von der Pütten, A. M., & Payr, S. (2011) · Applied Artificial Intelligence · doi:10.1080/08839514.2011.587153 · Cited by 99 · see citing works
Humans perceive warmth and competence in artificial intelligence
McKee, K. R., Bai, X., & Fiske, S. T. (2023) · iScience · doi:10.1016/j.isci.2023.107256 · Cited by 97 · see citing works
The renaissance of belief-based utility in economics
Loewenstein, G. F., & Molnár, A. (2018) · Nature Human Behaviour · doi:10.1038/s41562-018-0301-z · Cited by 96 · see citing works
Open Problems in Cooperative AI
Dafoe, A., Hughes, E., Bachrach, Y., Collins, T., McKee, K. R., Leibo, J. Z., Larson, K., & Graepel, T. (2020) · arXiv · doi:10.48550/arxiv.2012.08630 · Cited by 93 · see citing works
Looking at human–computer interface design: Effects of ethnicity in computer agents
Pratt, J. A., Hauser, K., Ugray, Z., & Patterson, O. V. (2007) · Interacting with Computers · doi:10.1016/j.intcom.2007.02.003 · Cited by 90 · see citing works
Uncertainty about the impact of social decisions increases prosocial behaviour
Kappes, A., Nussberger, A.-M., Faber, N. S., Kahane, G., Savulescu, J., & Crockett, M. J. (2018) · Nature Human Behaviour · doi:10.1038/s41562-018-0372-x · Cited by 79 · see citing works
March, C. (2021) · Journal of Economic Psychology · doi:10.1016/j.joep.2021.102426 · Cited by 72 · see citing works
The power to harm: AI assistants pave the way to unethical behavior
Gratch, J. S., & Fast, N. J. (2022) · Current Opinion in Psychology · doi:10.1016/j.copsyc.2022.101382 · Cited by 62 · see citing works
People Do Not Feel Guilty About Exploiting Machines
Melo, C. D., Marsella, S., & Gratch, J. S. (2016) · ACM Transactions on Computer-Human Interaction · doi:10.1145/2890495 · Cited by 59 · see citing works
Prosocial behavior toward machines
Nielsen, Y. A., Pfattheicher, S., & Keijsers, M. (2021) · Current Opinion in Psychology · doi:10.1016/j.copsyc.2021.08.004 · Cited by 58 · see citing works
Bellemare, C., Kröger, S., & van Soest, A. (2011) · Journal of Economic Behavior & Organization · doi:10.1016/j.jebo.2011.01.019 · Cited by 54 · see citing works
Social impact and governance of AI and neurotechnologies
Doya, K., Ema, A., Kitano, H., Sakagami, M., & Russell, S. (2022) · Neural Networks · doi:10.1016/j.neunet.2022.05.012 · Cited by 45 · see citing works
Bubbles in hybrid markets: How expectations about algorithmic trading affect human trading
Farjam, M., & Kirchkamp, O. (2017) · Journal of Economic Behavior & Organization · doi:10.1016/j.jebo.2017.11.011 · Cited by 44 · see citing works
Trust within human-machine collectives depends on the perceived consensus about cooperative norms
Makovi, K., Sargsyan, A., Li, W., Bonnefon, J.-F., & Rahwan, T. (2023) · Nature Communications · doi:10.1038/s41467-023-38592-5 · Cited by 36 · see citing works
Artificial intelligence, ethics, and intergenerational responsibility
Klockmann, V., von Schenk, A., & Villeval, M. C. (2022) · Journal of Economic Behavior & Organization · doi:10.1016/j.jebo.2022.09.010 · Cited by 31 · see citing works
Self-citationWe and It: An Interdisciplinary Review of the Experimental Evidence on Human-Machine InteractionPreprint
Chugunova, M., & Sele, D. (2020) · SSRN Electronic Journal · doi:10.3929/ethz-b-000442053 · Cited by 30 · see citing works
Turkers of the World Unite: Multilevel In-Group Bias Among Crowdworkers on Amazon Mechanical Turk
Almaatouq, A., Krafft, P. M., Dunham, Y. C., Rand, D. G., & Pentland, A. (2019) · Social Psychological and Personality Science · doi:10.1177/1948550619837002 · Cited by 26 · see citing works
Nielsen, Y. A., Thielmann, I., Zettler, I., & Pfattheicher, S. (2021) · Social Psychological and Personality Science · doi:10.1177/19485506211055622 · Cited by 21 · see citing works
Human—Android Interaction in the Near and Distant Future
Roese, N. J., & Amir, E. (2009) · Perspectives on Psychological Science · doi:10.1111/j.1745-6924.2009.01150.x · Cited by 14 · see citing works
Rent-seeking and competitive preferences
Cox, C. A. (2017) · Journal of Economic Psychology · doi:10.1016/j.joep.2017.02.002 · Cited by 14 · see citing works
Autonomous agents and human cultures in the trust–revenge game
Azaria, A., Richardson, A., & Rosenfeld, A. (2015) · Autonomous Agents and Multi-Agent Systems · doi:10.1007/s10458-015-9297-1 · Cited by 12 · see citing works
Giving Social Robots a Conversational Memory for Motivational Experience Sharing
Saravanan, A., Tsfasman, M., Neerincx, M. A., & Oertel, C. (2022) · IEEE RO-MAN · doi:10.1109/ro-man53752.2022.9900677 · Cited by 9 · see citing works
Artificial Intelligence, Ethics, and PivotalityPreprint
Klockmann, V., von Schenk, A., & Villeval, M. C. (2022) · SSRN Electronic Journal · doi:10.2139/ssrn.4003065 · Cited by 1 · see citing works
Self-citation
Flagged first, then by citations
Research works printed but not linked: 5
- Lin R., 2014, An International Journal, 30, 48
- McFadden D., 1980, The Journal of Business, 53
- McKee K. R., Bai X., Fiske S. T. (2022). Warmth and competence in human-agent cooperation. arXiv. https://doi.org/10.48550/arXiv.2201.1344 · DOI
- Russell S. J., 2019, Human compatible: AI and the problem of control
- Selten R., 1967, Beiträge zur experimentellen Wirtschaftforschung, 1
Cited by
Sorted by citations
The impact of generative artificial intelligence on socioeconomic inequalities and policy making
Capraro, V., Lentsch, A., Acemoglu, D., Akgun, S., Akhmedova, A., Bilancini, E., Bonnefon, J.-F., Brañas-Garza, P., Butera, L., Douglas, K. M., et al. (2024) · PNAS Nexus · doi:10.1093/pnasnexus/pgae191 · Cited by 371 · see citing works
You, Me, and the AI: The Role of Third‐Party Human Teammates for Trust Formation Toward AI Teammates
Erengin, T., Briker, R., & de Jong, S. B. (2025) · Journal of Organizational Behavior · doi:10.1002/job.2857 · Cited by 27 · see citing works
The impact of generative artificial intelligence on socioeconomic inequalities and policy makingPreprint
Capraro, V., Lentsch, A., Acemoğlu, D., Akgün, S., Akhmedova, A., Bilancini, E., Bonnefon, J., Brañas‐Garza, P., Butera, L., Douglas, K. M., et al. (2023) · doi:10.31234/osf.io/6fd2y · Cited by 36 · see citing works
Purcell, Z. A., Dong, M., Nußberger, A.-M., Köbis, N., & Jakesch, M. (2024) · British Journal of Psychology · doi:10.1111/bjop.12727 · Cited by 35 · see citing works
Self-citationThe Impact of Generative Artificial Intelligence on Socioeconomic Inequalities and Policy MakingPreprint
Capraro, V., Lentsch, A., Acemoğlu, D., Akgün, S., Akhmedova, A., Bilancini, E., Bonnefon, J., Brañas‐Garza, P., Butera, L., Douglas, K. M., et al. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4666103 · Cited by 13 · see citing works
Biswas, S., Erlei, A., & Gadiraju, U. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713201 · Cited by 12 · see citing works
The effects of social presence on cooperative trust with algorithms
Liefooghe, B., Min, E., & Aarts, H. (2023) · Scientific Reports · doi:10.1038/s41598-023-44354-6 · Cited by 12 · see citing works
Adverse reactions to the use of large language models in social interactions
Dvorak, F., Stumpf, R., Fehrler, S., & Fischbacher, U. (2025) · PNAS Nexus · doi:10.1093/pnasnexus/pgaf112 · Cited by 10 · see citing works
Human cooperation with artificial agents varies across countries
Karpus, J., Shirai, R., Verba, J. T., Schulte, R., Weigert, M., Bahrami, B., Watanabe, K., & Deroy, O. D. (2025) · Scientific Reports · doi:10.1038/s41598-025-92977-8 · Cited by 9 · see citing works
Social mindfulness and the SoMi Paradigm: A decade of research on low-cost prosociality
Doesum, N. J. V., Lange, P. A. M. V., Mischkowski, D., & van Dijk, E. (2025) · European Review of Social Psychology · doi:10.1080/10463283.2025.2457918 · Cited by 8 · see citing works
GPT-3.5 altruistic advice is sensitive to reciprocal concerns but not to strategic risk
Schmidt, E., Bonati, S., Köbis, N., & Soraperra, I. (2024) · Scientific Reports · doi:10.1038/s41598-024-73306-x · Cited by 6 · see citing works
Self-citationReputation-based reciprocity in human–bot and human–human networks
Harrell, A., & Traeger, M. L. (2025) · PNAS Nexus · doi:10.1093/pnasnexus/pgaf150 · Cited by 5 · see citing works
Evidence of spillovers from (non)cooperative human-bot to human-human interactions
Harrell, A., & Traeger, M. L. (2025) · iScience · doi:10.1016/j.isci.2025.113006 · Cited by 6 · see citing works
Leib, M., Köbis, N., & Soraperra, I. (2025) · Computers in Human Behavior · doi:10.1016/j.chb.2025.108709 · Cited by 5 · see citing works
Self-citationBecker, M., Jong, D. D., Briker, R., Mennens, K., Heller, J., Mahr, D., & Grewal, D. (2026) · The Leadership Quarterly · doi:10.1016/j.leaqua.2026.101952 · Cited by 4 · see citing works
Mackay, A. C. D., Zuo, L., & Kebe, I. A. (2026) · Behavioral Sciences · doi:10.3390/bs16040496 · Cited by 3 · see citing works
Ultimatum bargaining: Algorithms vs. Humans
Özkes, A., Hanaki, N., Vanderelst, D. F., & Willems, J. (2024) · Economics Letters · doi:10.1016/j.econlet.2024.111979 · Cited by 3 · see citing works
Wang, D., Liang, J. N., & Onwe, J. C. (2025) · Environmental Research Communications · doi:10.1088/2515-7620/adeb4f · Cited by 2 · see citing works
Makovi, K., Bonnefon, J., Oudah, M., Sargsyan, A., & Rahwan, T. (2025) · iScience · doi:10.1016/j.isci.2025.112833 · Cited by 2 · see citing works
Liu, W., Mark, I. G., Korn, C. W., & Rosenblau, G. (2025) · Journal of Adolescence · doi:10.1002/jad.70025 · Cited by 1 · see citing works
Comparing the audience effect of anthropomorphic robots and humans in economic games
Stinkeste, C., Dreber, A., Olofsson, J., & Skantze, G. (2025) · Computers in Human Behavior Artificial Humans · doi:10.1016/j.chbah.2025.100215 · Cited by 1 · see citing works
Wang, J.-X., Xue, Y., Zhang, Y.-T., Kuai, S., & Xie, X.-C. (2026) · Acta Psychologica · doi:10.1016/j.actpsy.2026.106752 · Cited by 1 · see citing works
Human and Algorithmic Cooperation: Parameters Matter!Preprint
von Schenk, A., Blonski, M., Klockmann, V., Eibelshäuser, S., & Hegemann, H. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4955708
Self-citationKardoš, P., Leidner, B., & Nawalkha, S. K. (2024) · The Journal of Psychology · doi:10.1080/00223980.2024.2437380
Over-monitoring under Transparency ParadoxPreprint
Lee, N. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5022072
Engel, C., Grossmann, M. R. P., & Ockenfels, A. (2025) · Experimental Economics · doi:10.1017/eec.2025.10038
Kou, Y., Zhou, Q., Wang, J., & Yu, N. X. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791151
Enhancing oneself vs. Undermining opponent in human-AI competitive contexts
Li, Y., & Zhou, X. (2026) · Social Sciences & Humanities Open · doi:10.1016/j.ssaho.2026.102960
Exploring trust in Artificial Intelligence (AI) systems: Insights from a repeated trust game
Jayasekara, D. N., Deng, R., Ho, J. Q., & Prissé, B. (2026) · Journal of Behavioral and Experimental Economics · doi:10.1016/j.socec.2026.102658 · Cited by 1 · see citing works
Informing users about LLMs' environmental costs encourages ethical consumption
Rabinovitch, H., Kashi, I., & Shalvi, S. (2026) · Computers in Human Behavior · doi:10.1016/j.chb.2026.109214
- Authors
- von Schenk, A., Klockmann, V., & Köbis, N.
- Affiliations
- Center for Humans and Machines, Max Planck Institute for Human Development
- University of Würzburg (DE)
- Max Planck Institute for Human Development (DE)
- Corresponding author
- Köbis, Nils
- Venue
- Perspectives on Psychological Science
- Publisher
- SAGE Publications
- Published
- 2023-09-26
- Volume
- 20
- Issue
- 1
- Pages
- 165-181
- ISSN
- 1745-6916 (ISSN-L)
- 1745-6924
- PMC
- PMC11720266
- OpenAlex
- W4387037619
- Cited by
- 30
- References
- 69
- Citations by year
- 2026: 7 · 2025: 14 · 2024: 7 · 2023: 2
- FWCI
- 7.49
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Locations
- Perspectives on Psychological Science (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- MPG.PuRe (Max Planck Society) (submittedVersion, other-oa)
- Topics
- Social Sciences › Safety Research › Experimental Behavioral Economics Studies
- Social Sciences › Psychology › Social Psychology › Social Robot Interaction and HRI
- Social Sciences › Sociology and Political Science › Social Power and Status Dynamics
- Keywords
- human-machine collaboration, reciprocity, self-serving bias, human-machine interaction
- MeSH terms
- Adult, Artificial Intelligence, Cooperative Behavior, Female, Humans, Male, Social Behavior, Young Adult
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.
Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.
Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.
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
