Pith. sign in

REVIEW 3 major objections 4 minor 85 references

My Advisor, Her AI and Me: Evidence from a Field Experiment on Human-AI Collaboration and Investment Decisions

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Human oversight of AI financial advice makes customers follow it more, and the effect comes through emotional trust rather than beliefs about quality.

desk verdict The field experiment's main effect—human-in-the-loop advice is followed more than pure AI advice—is credible and novel; the emotional-trust mechanism is not established because the mediator is measured after the outcome. read the letter →

arxiv 2506.03707 v1 pith:DYFHI6JC submitted 2025-06-04 econ.GN q-fin.EC

Cathy (Liu) Yang , Kevin Bauer , Xitong Li , Oliver Hinz This is my paper
classification econ.GNq-fin.EC
keywords human-in-the-loopAIadvicefieldexperimentadvice-takinginvestmentdecisionsemotionaltrustperipheralpersuasionalgorithmicaversion
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that who is seen to produce AI-based advice changes how much consumers use it, not just how good the advice is. In a field experiment with a savings bank, customers whose advice came from a banker working with an AI aligned their final investment decisions with the advice 15.5 percentage points more often than customers advised by the AI alone, with the gain concentrated in riskier loans. Because the underlying loans were real and payoffs were paid out, the extra advice-taking translated into materially higher consumer payoffs. The paper further claims that this effect is not driven by beliefs that human-AI advice is more accurate, but by the banker's presence acting as a peripheral persuasive cue that raises emotional trust, especially when uncertainty is high.

What carries the argument

The argument is carried by a two-stage field design: bankers first produced advice alone and then with an AI risk assessment, creating human-AI advice that demonstrably differs from the AI's, and customers were then randomly assigned to receive the AI advice, the human-AI advice, or the human-AI advice labeled as human. The pivotal measurement is FinalAlign, the alignment of a customer's final, incentive-compatible investment decision with the advised recommendation, analyzed with loan-by-advice and date fixed effects. On top of this, the mechanism is isolated by a moderated-mediation multilevel model in which decision uncertainty moderates the treatment's effect on emotional trust, and emotional trust mediates advice-taking.

What would settle it

Re-run the online study measuring emotional trust before the final investment decision, or in a separate session a day earlier; if the indirect effect of the human-AI label through emotional trust on advice alignment disappears or reverses when trust is measured before decisions, the peripheral-cue mechanism is not supported.

Watch

Extended reading notes

Core claim

Customers are more likely to follow investment advice when they believe a human banker had the final say over an AI's recommendation, compared with the same service delivered by the AI alone. Using ten real personal loans and incentivized decisions, the field experiment estimates an average increase of 15.5 percentage points in the share of final investment decisions that match the advice (beta 0.155, SE 0.046, p<0.001), rising to 21.3 points for riskier loans and near zero for safer ones. The same advice was presented as pure AI, as human-AI collaboration, or as human-only, and customers did not distinguish human-AI from human-only advice, which the authors read as evidence against a quality-based 'two advisors' explanation. An online replication with identical advice across conditions and measured prior beliefs confirms the pattern, and a moderated-mediation analysis shows the extra reliance is channeled through emotional trust rather than cognitive trust, disappointment, or perceived autonomy.

Load-bearing premise

The headline mechanism rests on emotional trust being measured after the investment decisions in the same session, so the trust scores may capture post-decision rationalization rather than the persuasion that caused the decisions.

Editorial extensions

If this is right

  • Regulation requiring humans in the loop changes downstream advice consumption, not just upstream quality.
  • Firms offering hybrid human-AI advisors can expect higher take-up of recommendations than pure AI, all else equal.
  • The consumption-side benefit appears in domains where the human oversight does not degrade advice quality; where it does, the same persuasion could amplify harm.
  • A human label has persuasive power even when consumers do not believe the human-AI advice is more accurate.
  • The effect is strongest exactly where decisions are riskiest and most uncertain.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the mechanism is affective trust, the persuasive boost of a human label may decay once the human-AI advice is experienced repeatedly, or may generalize to any human presence, even a non-expert, as long as the label is believed.
  • The design implies a welfare tradeoff: mandating human oversight improves outcomes only when the human's edits keep advice quality high; the same trust channel could make consumers worse off with low-quality human-AI advice, so quality monitoring is the load-bearing policy complement.
  • A testable extension is to vary only the final-sign-off cue, such as an unnamed 'human reviewer' versus a named banker, while holding advice content fixed, to separate accountability from social presence as the driver of the trust effect.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper studies whether human involvement in AI-based financial advice changes how downstream consumers respond to that advice. In a field experiment with a German savings bank, the authors first had human bankers produce advice in collaboration with a purpose-built AI model (Stage 1), then randomly assigned bank customers to receive either pure AI advice, human-AI collaborative advice, or human-AI advice labeled as human-only (Stage 2). The main estimate is that customers align their final investment decisions with the advice 15.5 percentage points more often in the human-AI condition than in the AI-only condition (Table 1), with the effect concentrated in riskier loans (Table 3) and translating into higher log-payoffs (Table 4). An online experiment with identical advice across conditions (n=87) replicates the main effect and provides a moderated mediation analysis suggesting that emotional trust mediates the effect under high decision uncertainty, which the authors interpret as peripheral-route persuasion. The paper also reports production-side results showing that bankers changed the AI advice but did not reduce its average quality.

Significance. If the main effect holds, the paper makes a valuable contribution by showing that human-in-the-loop AI systems shape consumer behavior on the consumption side, not just advice quality on the production side. The design has notable strengths: a real field setting with incentivized investment decisions, a pre-registered-style two-stage protocol (not explicitly stated as pre-registered), robustness to loan-by-advice fixed effects, logistic specifications, branch fixed effects, and inclusion of straight-liners, plus a model-free comparison on the four loans with identical advice across conditions (Figure 2). The online replication with identical advice across conditions is a particularly useful check because it removes the confound of advice differences, and the prior-belief measures help rule out one central-route story. However, the mechanism claim, which is a headline contribution, rests on a mediator measured after the outcome, making the causal interpretation of the peripheral-route conclusion fragile. The heterogeneity analysis also relies on a treatment-dependent risk split in the field data. The average treatment effect itself appears credible.

major comments (3)
  1. [Section 6.1.3 and Section 6.2.2] The emotional trust mediator is measured after participants made their final investment decisions: as described in Section 6.1.3, the trust items in Table A10 were collected 'before finishing the study,' i.e., after the incentivized choices. The moderated mediation result (indirect effect 0.040, 95% CI [0.005, 0.076], Section 6.2.2) therefore cannot distinguish the proposed causal chain (human-AI advice increases emotional trust, which increases advice-taking) from a post-decision rationalization account (participants who followed the advice under uncertainty later report higher emotional trust). The Table 5 interaction (Human-AI × PercUncertainDecision → EmoTrust, 3.554, SE 1.385) is equally compatible with both directions. Because the peripheral-route mechanism is a central contribution, this temporal-order problem is load-bearing. The paper should either present pre-decision measures of the mediator or substantially weaken the causal interpretation of the mediation analysis and present it as descriptive process evidence.
  2. [Section 5.4.2 and Table 3] The heterogeneity analysis splits the sample by the advisor's risk assessment, which differs between the Human-AI and AI-only conditions for six of the ten loans (Table A2). With loan-by-advice fixed effects in Equation (1), the treatment effect is identified only from the four loans with identical advice across conditions; for the six loans with differing advice, the treatment indicator is collinear with the fixed effect. Consequently, the field evidence for the riskiness heterogeneity is identified from a subset of loans whose composition within the risky/less-risky split is treatment-dependent. The authors should report the heterogeneity analysis restricted to the four same-advice loans, or at least clarify the identifying variation, to support the claim that the effect is concentrated in riskier loans.
  3. [Section 5.4.1 and Table 1] The reported average treatment effect of 0.155 (Table 1, Column 2) is identified only from the four loans with identical advice across conditions when loan-by-advice fixed effects are included. For the six loans where advice differs, the treatment indicator is perfectly explained by the loan-by-advice fixed effect, so those observations do not contribute to the estimate. The paper should state this explicitly when interpreting the economic magnitude, including the '44.92% increase in final payoff' claim in Section 5.5, because the payoff effect in Table 4 is also estimated off the same four-loan variation. This clarification is important for readers assessing the generalizability of the magnitude to the full set of loans.
minor comments (4)
  1. [Section 5.4.1] The text contains a duplicated word in the robustness-check description: 'non-linear logistic model specification specification.'
  2. [Table 4] The significance note for Table 4 reads '***p <0.01' for three stars, but the table also uses '****' for four stars; the note should list the four-star threshold consistently with Table 1 (e.g., '****p <0.001').
  3. [Section 6.1.3] The cognitive trust measure has a Cronbach's alpha of 0.41 and is reduced to two items after dropping the third; the resulting two-item correlation of 0.50 indicates a weak composite, and the paper should discuss the reliability of this measure more carefully or treat it as a secondary, exploratory construct.
  4. [Section 6.1.3] The attention-check exclusions are imbalanced across conditions (seven AI-only versus three Human-AI participants failed), so the paper could report the main online-experiment results without exclusions or with a treatment-by-attention interaction as a robustness check.

Circularity Check

0 steps flagged · score 0.0 of 10

No definitional circularity: the treatment effect is estimated from randomized field data, and the mechanism concern is a measurement-order validity issue rather than a self-referential derivation.

full rationale

The paper's central claim — that customers align final investment decisions with advice 15.5 percentage points more under Human-AI than AI-only — is estimated by OLS from a randomized field experiment (Table 1, Col. 2), not derived from the inputs that define it. The robustness on the four loans with identical advice across conditions (Table A8, Col. 4) and the online replication with identical advice (Table B1) show the effect is not an artifact of advice variation or a renamed fitted parameter. The mechanism claim (peripheral route via emotional trust) is the only part that invites concern: emotional trust was measured after the final investment decisions (Section 6.1.3), so the moderated mediation (Section 6.2.2) cannot rule out ex-post rationalization. But that is a temporal-ordering/mediation-identification problem, not a circularity in which a construct is defined in terms of the conclusion or an equation is equal to its input by construction. The ELM peripheral-route interpretation is applied after observing the riskiness heterogeneity (Section 5.4.2) and is explicitly framed as interpretation rather than a pre-registered derivation. Self-citations (You et al., 2022; Li et al., 2022) are used for prior literature and estimation method, not as load-bearing authority for the paper's uniqueness or for an unexamined ansatz. The paper also candidly lists limitations (no genuine human-only condition, 70% accuracy boundary, open confidence channel) that do not conceal missing support for the central estimate. Therefore no circular step meeting the quoted-reduction standard is present.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claims rest on the standard input of this literature: a fitted ML model that generates the advice content, behavioral assumptions about incentives and proxies (advisor risk assessment for uncertainty), a theoretical frame (ELM) chosen to interpret heterogeneity, and post hoc measurement choices (trust item selection, straight-liner exclusions) that are robustness-checked. The paper is transparent about most of these choices; the main unacknowledged fragility is the post-outcome measurement of the key mediator.

free parameters (4)
  • Risk-split threshold for risky versus less risky investments = Advised risk > 4 (moderate) on a 1-7 scale
    Section 5.4.2 and footnote 13: the median advised risk assessment of 4 defines the split; the footnote states this threshold was preferred because it yields a more balanced split, meaning it was chosen partly for data properties.
  • Cognitive and emotional trust scale items = Cognitive trust 3 to 2 items; emotional trust 3 to 2 items
    Section 6.1.3: items were dropped because of low Cronbach alpha (cognitive 0.41) and German translation ambiguity; final two-item correlations are 0.50 and 0.72. The key mediators are constructed from these post hoc item sets.
  • AI model architecture and hyperparameters = Neural network; exact configuration not reported
    Section 4.1: selected via five-fold cross-validation on the training set; the advice content customers receive depends on this model, whose performance (accuracy 0.73, AUC 0.71) is below the best published Lending Club models the authors cite.
  • Straight-liner exclusion rule = 7 of 137 customers excluded
    Section 5.3: post hoc quality screen removing customers who always made the same decision across loans; a robustness check including these customers is reported in Table A8 Column 3, so the exclusion is transparent but still data-dependent.
assumptions (5)
  • domain assumption The advised risk assessment is a valid proxy for customer decision uncertainty.
    Section 5.4.2 footnote 13: the authors use the advisor's risk assessment as a proxy for uncertainty and note the two constructs are distinct but positively correlated.
  • domain assumption The elaboration likelihood model applies to retail investment advice and predicts peripheral-route dominance under uncertainty.
    Section 2.2: ELM is imported to generate the central versus peripheral mechanism predictions; the heterogeneity result is interpreted through ELM rather than compared with rival models.
  • domain assumption The Human-only condition, which presents identical human-AI advice without disclosing AI use, isolates the human-involvement cue from complementarity beliefs.
    Sections 5.1 and 5.4.2: the inference that central-route complementarity beliefs do not drive the effect rests on the null difference between Human-AI and Human-only; the label change is assumed to manipulate only the disclosed source.
  • domain assumption The lottery-incentive design makes investment decisions consequential enough to measure advice-taking.
    Section 5.2: customers were told a minimum of 10% of participants would have one of twenty decisions realized; the authors treat final payoff averages as evidence of engagement.
  • standard math Linear probability models with fixed effects recover the average treatment effect in this panel.
    Section 5.4.1: the incidental parameter justification and the boundary check (99.88% of predicted probabilities within [0,1]) support OLS on the binary outcome; a logistic robustness check is also reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of My Advisor, Her AI and Me: Evidence from a Field Experiment on Human-AI Collaboration and Investment Decisions." pith.science (2026). https://pith.science/paper/DYFHI6JC

@misc{pith2026250603707,
  author       = {Pith},
  title        = {Pith review of: My Advisor, Her AI and Me: Evidence from a Field Experiment on Human-AI Collaboration and Investment Decisions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DYFHI6JC}},
  note         = {Machine review of arXiv:2506.03707}
}
read the original abstract

Amid ongoing policy and managerial debates on keeping humans in the loop of AI decision-making, we investigate whether human involvement in AI-based service production benefits downstream consumers. Partnering with a large savings bank in Europe, we produced pure AI and human-AI collaborative investment advice, passed it to customers, and examined their advice-taking in a field experiment. On the production side, contrary to concerns that humans might inefficiently override AI output, we find that giving a human banker the final say over AI-generated financial advice does not compromise its quality. More importantly, on the consumption side, customers are more likely to follow investment advice from the human-AI collaboration compared to pure AI, especially when facing riskier decisions. In our setting, this increased reliance leads to higher material welfare for consumers. Additional analyses from the field experiment and an online experiment show that the persuasive power of human-AI advice cannot be explained by consumers' beliefs about enhanced advice quality due to human-AI complementarities. Instead, the benefit stems from human involvement acting as a peripheral cue that increases the advice's affective appeal. Our findings suggest that regulations and guidelines should adopt a consumer-centered approach by fostering service environments in which humans and AI systems can collaborate to improve consumer outcomes. These insights are relevant for managers designing AI-based services and for policymakers advocating for human oversight in AI systems.

Figures

Figures reproduced from arXiv: 2506.03707 by the authors.

Figure 2
Figure 2. Model-free evidence when the Human-AI and AI advisor offered the same investment advice [PITH_FULL_IMAGE:figures/full_fig_p022_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

85 extracted references · 78 canonical work pages

  1. [1]

    M., Pfeuffer, N., Carl, K

    Abdel-Karim, B. M., Pfeuffer, N., Carl, K. V., and Hinz, O. (2023). How ai-based systems can induce reflections: The case of ai-augmented diagnostic work. MIS Quarterly , 47(4):1395--1424

    2023

  2. [2]

    and Restrepo, P

    Acemoglu, D. and Restrepo, P. (2019). Automation and new tasks: How technology displaces and reinstates labor. Journal of Economic Perspectives , 33(2):3--30

    2019

  3. [3]

    S., and Goldfarb, A

    Agrawal, A., Gans, J. S., and Goldfarb, A. (2024). Prediction machines, insurance, and protection: An alternative perspective on ai’s role in production. Journal of the Japanese and International Economies , 72:101307

    2024

  4. [4]

    Angrist, J. D. and Pischke, J.-S. (2009). Mostly harmless econometrics: An empiricist's companion . Princeton university press

    2009

  5. [5]

    Asch, S. E. (1955). Opinions and social pressure. Scientific American , 193(5):31--35

    1955

  6. [6]

    Avramov, D., Cheng, S., and Metzker, L. (2023). Machine learning vs. economic restrictions: Evidence from stock return predictability. Management Science , 69(5):2587--2619

    work page 2023

  7. [7]

    Awad, E., Dsouza, S., Kim, R., Schulz, J., Henrich, J., Shariff, A., Bonnefon, J.-F., and Rahwan, I. (2018). The moral machine experiment. Nature , 563(7729):59--64

    work page 2018

  8. [8]

    Bauer, K., Heigl, R., Hinz, O., and Kosfeld, M. (2024a). Feedback loops in machine learning: A study on the interplay of continuous updating and human discrimination. Journal of the Association for Information Systems , 25(4):804--866

    work page 2024

Show all 85 references
  1. [9]

    Bauer, K., Jussupow, E., Heigl, R., Vogt, B., and Hinz, O. (2024b). All just in your head? unraveling the side effects of generative ai disclosure in creative task (february 29, 2024). Working paper, SSRN. Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4782554

    2024

  2. [10]

    Bauer, K., von Zahn, M., and Hinz, O. (2023). Expl (ai) ned: The impact of explainable artificial intelligence on users’ information processing. Information systems research , 34(4):1582--1602

    2023

  3. [11]

    Bergstein, B. (2020). What ai still can’t do. MIT Technology Review , 123(2):1--7

    2020

  4. [12]

    Berk, R. (2017). An impact assessment of machine learning risk forecasts on parole board decisions and recidivism. Journal of Experimental Criminology , 13(2):193--216

    2017

  5. [13]

    and Dalal, R

    Bonaccio, S. and Dalal, R. S. (2006). Advice taking and decision-making: An integrative literature review, and implications for the organizational sciences. Organizational Behavior and Human Decision Processes , 101(2):127--151

    2006

  6. [14]

    Brynjolfsson, E., Mitchell, T., and Rock, D. (2018). What can machines learn, and what does it mean for occupations and the economy? AEA Papers and Proceedings , 108:43--47

    2018

  7. [15]

    P., Zetzsche, D

    Buckley, R. P., Zetzsche, D. A., Arner, D. W., and Tang, B. W. (2021). Regulating artificial intelligence in finance: Putting the human in the loop. The Sydney Law Review , 43(1):43--81

    2021

  8. [16]

    Cao, S., Jiang, W., Wang, J., and Yang, B. (2024). From man vs. machine to man+ machine: The art and ai of stock analyses. Journal of Financial Economics , 160:103910

    2024

  9. [17]

    W., and Lehmann, D

    Castelo, N., Bos, M. W., and Lehmann, D. R. (2019). Task-dependent algorithm aversion. Journal of Marketing Research , 56(5):809--825

    2019

  10. [18]

    Chang, A.-H., Yang, L.-K., Tsaih, R.-H., and Lin, S.-K. (2022). Machine learning and artificial neural networks to construct p2p lending credit-scoring model: A case using lending club data. Quantitative Finance and Economics , 6(2):303--325

    2022

  11. [19]

    Commission, E. (2019). Communication from the commission to the european parliament, the council, the european economic and social committee and the committee of the regions - building trust in human centric artificial intelligence (com(2019)168). https://digital-strategy.ec.e...

    2019

  12. [20]

    C., Woo, C

    Cooper, A. C., Woo, C. Y., and Dunkelberg, W. C. (1988). Entrepreneurs' perceived chances for success. Journal of Business Venturing , 3(2):97--108

    1988

  13. [21]

    and Henshaw, J

    Costa, P. and Henshaw, J. E. (2022). Quantifying the investor’s view on the value of human and robo-advice. Vanguard Research Report , (1)

    2022

  14. [22]

    and Tucker, C

    Cowgill, B. and Tucker, C. E. (2020). Algorithmic fairness and economics. Working paper, Columbia Business School Research Paper. Available at SSRN: https://ssrn.com/abstract=3361280 or http://dx.doi.org/10.2139/ssrn.3361280

    2020 doi

  15. [23]

    Curran, P. G. (2016). Methods for the detection of carelessly invalid responses in survey data. Journal of Experimental Social Psychology , 66:4--19

    2016

  16. [24]

    Dalal, R. S. and Bonaccio, S. (2010). What types of advice do decision-makers prefer? Organizational Behavior and Human Decision Processes , 112(1):11--23

    2010

  17. [25]

    Darda, K. M. and Cross, E. S. (2023). The computer, a choreographer? aesthetic responses to randomly-generated dance choreography by a computer. Heliyon , 9(1)

    2023

  18. [26]

    Dell’Acqua, F. (2022). Falling asleep at the wheel: Human/ai collaboration in a field experiment on hr recruiters. Technical report, Working paper

    2022

  19. [27]

    J., Simmons, J

    Dietvorst, B. J., Simmons, J. P., and Massey, C. (2015). Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General , 144(1):114

    2015

  20. [28]

    Dohmen, T., Falk, A., Huffman, D., and Sunde, U. (2012). The intergenerational transmission of risk and trust attitudes. The Review of Economic Studies , 79(2):645--677

    2012

  21. [29]

    Dong, H., Liu, R., and Tham, A. W. (2024). Accuracy comparison between five machine learning algorithms for financial risk evaluation. Journal of Risk and Financial Management , 17(2):50

    2024

  22. [30]

    EU (2024). Proposal for a regulation of the european parliament and of the council laying down harmonised rules on artificial intelligence (artificial intelligence act) and amending certain union legislative acts. https://data.consilium.europa.eu/doc/document/ST-5662-2024-INIT...

    2024

  23. [31]

    F., Saqlain, M

    Faisal, M. F., Saqlain, M. N. U., Bhuiyan, M. A. S., Miraz, M. H., and Patwary, M. J. (2021). Credit approval system using machine learning: Challenges and future directions. In 2021 International Conference on Computing, Networking, Telecommunications & Engineering Sciences A...

    2021

  24. [32]

    R., von Krogh, G., and Zhang, C

    Feuerriegel, S., Shrestha, Y. R., von Krogh, G., and Zhang, C. (2022). Bringing artificial intelligence to business management. Nature Machine Intelligence , 4(7):611--613

    2022

  25. [33]

    and Wright, P

    Friestad, M. and Wright, P. (1994). The persuasion knowledge model: How people cope with persuasion attempts. Journal of Consumer Research , 21(1):1--31

    1994

  26. [34]

    F \"u gener, A., Grahl, J., Gupta, A., and Ketter, W. (2021). Will humans-in-the-loop become borgs? merits and pitfalls of working with ai. MIS Quarterly , 45(3):1527--1556

    2021

  27. [35]

    F \"u gener, A., Grahl, J., Gupta, A., and Ketter, W. (2022). Cognitive challenges in human--artificial intelligence collaboration: Investigating the path toward productive delegation. Information Systems Research , 33(2):678--696

    2022

  28. [36]

    A., Kanjee, Z., Parsons, A

    Goh, E., Gallo, R., Hom, J., Strong, E., Weng, Y., Kerman, H., Cool, J. A., Kanjee, Z., Parsons, A. S., Ahuja, N., et al. (2024). Large language model influence on diagnostic reasoning: a randomized clinical trial. JAMA Network Open , 7(10):e2440969--e2440969

    2024

  29. [37]

    Greene, W. (2004). Fixed effects and bias due to the incidental parameters problem in the tobit model. Econometric reviews , 23(2):125--147

    2004

  30. [38]

    and Fischer, I

    Harvey, N. and Fischer, I. (1997). Taking advice: Accepting help, improving judgment, and sharing responsibility. Organizational Behavior and Human Decision Processes , 70(2):117--133

    1997

  31. [39]

    Hayes, A. F. (2017). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach . Guilford publications

    2017

  32. [40]

    Hermosilla, M., Guti \'e rrez-Navratil, F., and Prieto-Rodriguez, J. (2018). Can emerging markets tilt global product design? impacts of chinese colorism on hollywood castings. Marketing Science , 37(3):356--381

    2018

  33. [41]

    M., Gandhi, L., and Blaser, T

    Kahneman, D., Rosenfield, A. M., Gandhi, L., and Blaser, T. (2016). Noise: How to overcome the high, hidden cost of inconsistent decision making. Harvard business review , 94(10):38--46

    2016

  34. [42]

    C., Valentine, M

    Kellogg, K. C., Valentine, M. A., and Christin, A. (2020). Algorithms at work: The new contested terrain of control. Academy of management annals , 14(1):366--410

    2020

  35. [43]

    Kingston, J. K. (2016). Artificial intelligence and legal liability. In International Conference on Innovative Techniques and Applications of Artificial Intelligence XXIV , number 33, pages 269--279. Springer

    2016

  36. [44]

    F., DiDio, L., and Gagn \'e , M

    Koestner, R., Gingras, I., Abutaa, R., Losier, G. F., DiDio, L., and Gagn \'e , M. (1999). To follow expert advice when making a decision: An examination of reactive versus reflective autonomy. Journal of Personality , 67(5):851--872

    1999

  37. [45]

    Komiak, S. Y. and Benbasat, I. (2006). The effects of personalization and familiarity on trust and adoption of recommendation agents. MIS Quarterly , 30(4):941--960

    2006

  38. [46]

    Kumar, V., Natarajan, S., Keerthana, S., Chinmayi, K., and Lakshmi, N. (2016). Credit risk analysis in peer-to-peer lending system. In 2016 IEEE international conference on knowledge engineering and applications (ICKEA) , pages 193--196. IEEE

    2016

  39. [47]

    Lebovitz, S., Lifshitz-Assaf, H., and Levina, N. (2022). To engage or not to engage with ai for critical judgments: How professionals deal with opacity when using ai for medical diagnosis. Organization Science , 33(1):126--148

    2022

  40. [48]

    Lessmann, S., Baesens, B., Seow, H.-V., and Thomas, L. C. (2015). Benchmarking state-of-the-art classification algorithms for credit scoring: An update of research. European Journal of Operational Research , 247(1):124--136

    2015

  41. [49]

    Li, X., Grahl, J., and Hinz, O. (2022). How do recommender systems lead to consumer purchases? a causal mediation analysis of a field experiment. Information Systems Research , 33(2):620--637

    2022

  42. [50]

    F., Weber, E

    Loewenstein, G. F., Weber, E. U., Hsee, C. K., and Welch, N. (2001). Risk as feelings. Psychological Bulletin , 127(2):267

    2001

  43. [51]

    Longoni, C., Bonezzi, A., and Morewedge, C. K. (2019). Resistance to medical artificial intelligence. Journal of Consumer Research , 46(4):629--650

    2019

  44. [52]

    and Zhang, Y

    Lu, T. and Zhang, Y. (2024). 1+1> 2? information, humans, and machines. Information Systems Research , forthcoming

    2024

  45. [53]

    MacCrimmon, K. R. and Wehrung, D. A. (1990). Characteristics of risk taking executives. Management science , 36(4):422--435

    1990

  46. [54]

    and Aksakalli, V

    Malekipirbazari, M. and Aksakalli, V. (2015). Risk assessment in social lending via random forests. Expert Systems with Applications , 42(10):4621--4631

    2015

  47. [55]

    March, J. G. and Shapira, Z. (1987). Managerial perspectives on risk and risk taking. Management Science , 33(11):1404--1418

    1987

  48. [56]

    A I in F inance M arket S ize, S hare, G rowth R eport - 2030 --- marketsandmarkets.com

    MarketsandMarkets (2025). A I in F inance M arket S ize, S hare, G rowth R eport - 2030 --- marketsandmarkets.com. https://www.marketsandmarkets.com/Market-Reports/ai-in-finance-market-90552286.html. [Accessed 03-03-2025]

    2025

  49. [57]

    H., Choudhury, V., and Kacmar, C

    McKnight, D. H., Choudhury, V., and Kacmar, C. (2002). Developing and validating trust measures for e-commerce: An integrative typology. Information Systems Research , 13(3):334--359

    2002

  50. [58]

    Meade, A. W. and Craig, S. B. (2012). Identifying careless responses in survey data. Psychological Methods , 17(3):437

    2012

  51. [59]

    Millet, K., Buehler, F., Du, G., and Kokkoris, M. D. (2023). Defending humankind: Anthropocentric bias in the appreciation of ai art. Computers in Human Behavior , 143:107707

    2023

  52. [60]

    and Metilda, M

    Mishra, K. and Metilda, M. J. (2015). A study on the impact of investment experience, gender, and level of education on overconfidence and self-attribution bias. IIMB Management Review , 27(4):228--239

    2015

  53. [61]

    O nkal, D., Goodwin, P., Thomson, M., G \

    \"O nkal, D., Goodwin, P., Thomson, M., G \"o n \"u l, S., and Pollock, A. (2009). The relative influence of advice from human experts and statistical methods on forecast adjustments. Journal of Behavioral Decision Making , 22(4):390--409

    2009

  54. [62]

    Petty, R. E. and Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion. In Communication and Persuasion , pages 1--24. Springer

    1986

  55. [63]

    M., Schramowski, P., Bucher, A

    Pfeuffer, N., Baum, L., Stammer, W., Abdel-Karim, B. M., Schramowski, P., Bucher, A. M., H \"u gel, C., Rohde, G., Kersting, K., and Hinz, O. (2023). Explanatory interactive machine learning: Establishing an action design research process for machine learning projects. Busines...

    2023

  56. [64]

    J., Zhang, Z., and Zyphur, M

    Preacher, K. J., Zhang, Z., and Zyphur, M. J. (2016). Multilevel structural equation models for assessing moderation within and across levels of analysis. Psychological Methods , 21(2):189

    2016

  57. [65]

    W., Christakis, N

    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). Machine behaviour. Nature , 568(7753):477--486

    2019

  58. [66]

    Rohaan, D., Topan, E., and Groothuis-Oudshoorn, C. G. (2022). Using supervised machine learning for b2b sales forecasting: A case study of spare parts sales forecasting at an after-sales service provider. Expert Systems with Applications , 188:115925

    2022

  59. [67]

    P., Pavan, T

    Singh, R., Ayyar, M. P., Pavan, T. V. S., Gosain, S., and Shah, R. R. (2019). Automating car insurance claims using deep learning techniques. In 2019 IEEE Fifth International Conference on Multimedia Big Data (BigMM) , pages 199--207. IEEE

    2019

  60. [68]

    Slovic, P. (1987). Perception of risk. Science , 236(4799):280--285

    1987

  61. [69]

    Sniezek, J. A. and Buckley, T. (1995). Cueing and cognitive conflict in judge-advisor decision making. Organizational Behavior and Human Decision Processes , 62(2):159--174

    1995

  62. [70]

    Sniezek, J. A. and Van Swol, L. M. (2001). Trust, confidence, and expertise in a judge-advisor system. Organizational Behavior and Human Decision Processes , 84(2):288--307

    2001

  63. [71]

    F., and Perrmann-Graham, J

    Steffel, M., Williams, E. F., and Perrmann-Graham, J. (2016). Passing the buck: Delegating choices to others to avoid responsibility and blame. Organizational Behavior and Human Decision Processes , 135:32--44

    2016

  64. [72]

    H., Lewicki, R

    Strickland, L. H., Lewicki, R. J., and Katz, A. M. (1966). Temporal orientation and perceived control as determinants of risk-taking. Journal of Experimental Social Psychology , 2(2):143--151

    1966

  65. [73]

    J., Hu, H., and Van Mieghem, J

    Sun, J., Zhang, D. J., Hu, H., and Van Mieghem, J. A. (2022). Predicting human discretion to adjust algorithmic prescription: A large-scale field experiment in warehouse operations. Management Science , 68(2):846--865

    2022

  66. [74]

    Taudien, A., Fuegener, A., Gupta, A., and Ketter, W. (2022). Calibrating users’ mental models for delegation to AI . International Conference of Information Systems (ICIS)

    2022

  67. [75]

    and Kahneman, D

    Tversky, A. and Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science , 185(4157):1124--1131

    1974

  68. [76]

    von Zahn, M., Bauer, K., Mihale-Wilson, C., Jagow, J., Speicher, M., and Hinz, O. (2024). Smart green nudging: Reducing product returns through digital footprints and causal machine learning. Marketing Science , forthcoming

    2024

  69. [77]

    and Spindler, M

    Wasserbacher, H. and Spindler, M. (2022). Machine learning for financial forecasting, planning and analysis: recent developments and pitfalls. Digital Finance , 4(1):63--88

    2022

  70. [78]

    U., Blais, A.-R., and Betz, N

    Weber, E. U., Blais, A.-R., and Betz, N. E. (2002). A domain-specific risk-attitude scale: Measuring risk perceptions and risk behaviors. Journal of Behavioral Decision Making , 15(4):263--290

    2002

  71. [79]

    Weber, E. U. and Milliman, R. A. (1997). Perceived risk attitudes: Relating risk perception to risky choice. Management Science , 43(2):123--144

    1997

  72. [80]

    Wooldridge, J. M. (2010). Econometric analysis of cross section and panel data . MIT press

    2010

  73. [81]

    Yaniv, I. (2004). Receiving other people’s advice: Influence and benefit. Organizational Behavior and Human Decision Processes , 93(1):1--13

    2004

  74. [82]

    Yin, M., Wortman Vaughan, J., and Wallach, H. (2019). Understanding the effect of accuracy on trust in machine learning models. In Proceedings of the 2019 chi conference on human factors in computing systems , pages 1--12

    2019

  75. [83]

    L., and Li, X

    You, S., Yang, C. L., and Li, X. (2022). Algorithmic versus human advice: Does presenting prediction performance matter for algorithm appreciation? Journal of Management Information Systems , 39(2):336--365

    2022

  76. [84]

    , " * write output.state after.block = add.period write newline

    ENTRY address author booktitle chapter doi edition editor eid howpublished institution journal key month note number organization pages publisher school series title type url volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence afte...

  77. [85]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.