Pith. sign in

REVIEW 5 major objections 6 minor 35 references

Right vs. Right: Can LLMs Make Tough Choices?

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read LLMs show consistent, pronounced value preferences in right-vs-right dilemmas, favoring truth over loyalty in 93% of cases and community over the individual in 72%.

desk verdict A useful, well-scoped empirical map of LLM moral preferences on right-vs-right dilemmas, but the headline percentages rest on an LLM-generated dataset that has not been human-validated or released, so the results should be read as provisional. read the letter →

arxiv 2412.19926 v1 pith:ST6I3WSL submitted 2024-12-27 cs.CL

Jiaqing Yuan , Pradeep K. Murukannaiah , Munindar P. Singh This is my paper · ORCID
classification cs.CL
keywords ethicaldilemmaslargelanguagemodelsmoralvaluepreferencesalignmentKidder'sframeworkdeontologyconsequentialismpromptsensitivity
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

This paper asks what large language models do when every option is morally defensible — dilemmas of right versus right, not right versus wrong. Drawing on Kidder's four recurring value conflicts (truth vs. loyalty, individual vs. community, short-term vs. long-term, justice vs. mercy), the authors build a corpus of 1,730 dilemmas and test 20 models from six families. On scenarios where a model consistently agrees with its own answers, it favors truth over loyalty in 93.48% of cases, long-term over short-term interests in 83.69%, community over the individual in 72.37%, and justice over mercy in 68.49%. Larger models hold these choices even when told the outcomes are negative, and explicit statements of a preferred value steer models far more reliably than example cases. Because the same models agree with themselves only 60–80% of the time across different phrasings of a dilemma, these preferences sit atop a measurable sensitivity to wording — and that combination matters for any system that delegates tough choices to an LLM.

What carries the argument

The engine of the study is Kidder's four-paradigm taxonomy of right-vs-right dilemmas, which supplies the four value conflicts — truth vs. loyalty, individual vs. community, short-term vs. long-term, justice vs. mercy — that every scenario must instantiate. On top of it sits a two-step generation pipeline: gpt-3.5-turbo first lists domains where such dilemmas arise, then gpt-3.5-turbo and claude-3-sonnet each write scenarios with two actions, one aligned with each value, plus positive and negative consequences for each action; the only filtering is format compliance and mention of both value words. The measurement apparatus is twelve prompt templates: four phrasings (direct, direct-reverse, compare, compare-reverse) probe the baseline preference and the model's symbol binding; four consequence permutations probe whether outcomes move the choice; and explicit-preference plus few-shot templates (one, five, ten examples) probe steerability. The analytical keystone is unanimity filtering: only scenarios where a model answers identically under all four baseline phrasings contribute to its preference statistics, an attempt to separate a stable moral lean from prompt noise.

What would settle it

Have independent human raters classify a random sample of the 1,730 dilemmas: which value each action instantiates, and whether the scenario truly poses a right-vs-right conflict. If raters frequently disagree with the generator's intended assignments, or if the measured preference rates (truth over loyalty at 93.48%, for example) fail to replicate on the subset of scenarios that pass human validation, then the reported percentages are artifacts of the generation pipeline rather than stable properties of LLM moral reasoning.

Watch

Extended reading notes

Core claim

The central discovery is that LLMs show strong, repeatable value preferences when every option in a dilemma is morally defensible, and that those preferences survive changes in wording, option order, and even stated consequences. Restricting the analysis to twelve models that passed a self-consistency threshold, and to scenarios where a model gave the same answer under all four baseline prompt formulations, the authors report that truth is chosen over loyalty in 93.48% of cases, long-term over short-term interests in 83.69%, community over the individual in 72.37%, and justice over mercy in 68.49%. A second finding is scale-dependent: the largest, newest models flip their choice in only 5–10% of scenarios when consequences are reversed — a deontological stance, adhering to the initial rule regardless of outcomes — while smaller models flip in up to roughly 32% of scenarios. A third concerns steerability: an explicit statement that one value outranks another reverses the model's baseline choice in up to about 85% of scenarios, whereas demonstrating the preference through one to ten examples reverses it in at most about 44%. A fourth is a floor of fragility: even the best models agree with themselves only 60–80% of the time across four phrasings of the same dilemma, so the measured preferences sit on top of substantial prompt-formulation sensitivity.

Load-bearing premise

The load-bearing premise is that the AI-generated scenarios and their paired actions really instantiate the intended value conflict — that 'Action A' genuinely expresses truth and 'Action B' genuinely expresses loyalty in each of the 1,730 dilemmas — since the corpus was filtered only for format compliance and for mentioning the two value words, with no human validation reported.

Editorial extensions

If this is right

  • Deploying an LLM where value trade-offs are routine — health care, journalism, human resources, governance — means inheriting a specific moral lean (truth over loyalty, community over individual, long-term over short-term) that the model will not necessarily surface or justify on its own.
  • Operators who want a particular ethical stance should state it explicitly in the prompt: explicit value guidelines reversed models' choices in up to about 85% of scenarios, while in-context examples managed at most about 44%.
  • Larger models' deontological stability means outcome information will not talk them out of an initial moral choice, so applications that want consequence-sensitive decisions should not rely on large models' default behavior.
  • Self-agreement of only 60–80% across prompt phrasings means any single-prompt answer to an ethical question carries a formulation-dependent noise floor; agreement checks across phrasings are a cheap way to flag unreliable answers.
  • The justice-vs-mercy pair shows the thinnest consensus (68.49%) and the most variation across both consequence and preference manipulations, marking it as the most context-dependent and steerable of the four conflicts.

Reading between the lines

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

  • Because the 1,730 scenarios were generated by just two models with no human validation of the value assignments, the measured preference rates could partly reflect the generation pipeline rather than LLM moral reasoning; a human-rater study of a random sample would separate the two.
  • The unanimity filter keeps only the clearest dilemmas for each model, so the reported percentages likely overstate how often these preferences appear in messy real-world cases; computing the same rates on the full, non-unanimous responses would give a lower-bound estimate.
  • Since the two generator models were built largely on English-heavy, Western-oriented corpora, the truth-over-loyalty dominance may encode a particular training distribution; regenerating the corpus with models trained on other cultural data could show whether the profile is universal or inherited.
  • The explicit-beats-examples result suggests a practical alignment recipe — state the value hierarchy plainly — but also invites a follow-up question the paper does not ask: whether models can detect when a stated preference and a scenario's content pull in opposite directions.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper presents a descriptive study of how large language models behave when confronted with ethical dilemmas in which both options are presented as morally defensible. The authors construct a corpus of 1,730 dilemmas based on Kidder's four value-conflict pairs (Truth vs. Loyalty, Individual vs. Community, Short-Term vs. Long-Term, Justice vs. Mercy) using two LLM generators, then evaluate 20 LLMs across six families under multiple prompt formulations. The main reported findings are that LLMs prefer truth over loyalty, long-term over short-term benefits, and community over individual, with a weaker preference in the Justice/Mercy pair; that larger models are less likely to change their choices when consequences are added; and that explicitly stated value preferences steer models more effectively than in-context examples. The paper also reports that LLMs vary substantially in their sensitivity to prompt formulation. The study is broad in scope and addresses an underexplored question, but the central empirical claims rest on an unvalidated, unreleased LLM-generated dataset, and several reporting gaps make the robustness of the headline percentages difficult to assess.

Significance. If the dataset-construction assumptions hold, this would be a valuable descriptive contribution to machine ethics and value alignment: it provides a systematic map of LLM preferences across a well-known ethical taxonomy, covers a much larger scenario space than prior dilemma-based studies, and includes useful comparisons of prompt families with reversal controls. The finding that explicit value statements outperform few-shot demonstrations is practically relevant for prompt-design and alignment work. The paper is honest about some limitations, including prompt sensitivity and the simplified nature of its ethical framework. However, the study does not ship code or data, and its empirical claims currently depend on an LLM-generated corpus with no human validation and no reported inter-annotator agreement, which is a serious correctness risk for every headline percentage. The lack of statistical inference and the inconsistent reporting of the Justice/Mercy direction further weaken the current presentation.

major comments (5)
  1. [Section 3.2] The dataset-construction step is load-bearing for every headline result, and it is not validated. The 1,730 scenarios, the two actions, the assignment of each action to one side of the value pair, and the positive/negative consequences are all generated by gpt-3.5-turbo and claude-3-sonnet, and the only filters applied are format compliance and whether the explanation mentions the two value names. There is no human validation that Action 1 actually instantiates value 1, that Action 2 instantiates value 2, that the scenario is a genuine right-vs-right dilemma rather than a temptation scenario, or that the generated consequences have the intended valence. Because the generation prompt asks the model to label actions by the very values under test, the aggregate preference rates in Figure 2 and the consequence-flipping rates in Figure 3 can reflect the generators' biases or prompt artifacts rather than properties of the evaluated LLMs. The circularity is compounded by the fact that GPT-3.5 and Claude-3-Sonnet are themselves among the 20 evaluated models. The dataset is not released, so these concerns cannot be checked post hoc. Please report a human validation study on a representative sample, state the resulting agreement rates, and release the dataset or provide a clear availability statement.
  2. [Section 5.2, Figure 2] The headline preference rates (e.g., 'Truth is overwhelmingly preferred over Loyalty, with an average selection rate of 93.48%') are computed only on scenarios in which a model gives unanimous answers across all four prompt variants, after excluding all models whose all-four-prompt agreement is below that of Llama-3-8B (47.4%). This removes Claude-3-Sonnet, Mixtral-8x22B, Llama-2 (7B, 13B), Qwen2 (0.5B, 1.5B, 7B), and Yi-1.5 (6B, 9B) from Figure 2, so the cross-model comparison covers only 12 of the 20 evaluated models. The manuscript does not report how many scenarios survive the unanimity filter per model and per value pair, nor does it quantify the resulting selection bias. Restricting to unanimously answered scenarios can amplify any systematic bias in the generated dataset, because a generator that makes one action consistently more attractive will also produce more unanimous answers. Please report the number of retained scenarios per model/pair and show that the main conclusions are robust when the analysis is repeated on all scenarios, or justify the unanimity restriction as the primary analysis.
  3. [Sections 5.2 and 6; Figure 2] The direction of the Justice-vs-Mercy preference is reported inconsistently. Section 5.2 states 'Mercy is chosen over Justice 68.49% of the time,' and the first block of numbers in Figure 2 is consistent with that reading. However, Section 6 (Summary of Findings) states 'an average of 68.49% favoring Justice.' The abstract omits the Justice/Mercy pair entirely. Since this is one of the four central descriptive findings, the discrepancy must be corrected, and the abstract should state the direction if the finding is included.
  4. [Section 5.3, Figure 3] The claim that 'larger and more advanced models have a greater inclination to adopt the deontological principle' is stronger than the data shown. The flip rates in Figure 3 are not monotonic in model size or recency: Yi-1.5-34B flips 16.5%, Mixtral-8x7B 20.7%, and Llama-2-70B 12.3%, while Claude-3-Haiku flips 18.7%, so the 'larger' generalization is supported only by a loose trend among a subset of models. No confidence intervals or significance tests are reported for these flip rates or for the other headline percentages. Please either temper the claim to 'some of the largest models in this sample' or provide a statistical analysis that supports the size/recency generalization.
  5. [Section 5, Table 9] The handling of invalid responses is not reported. Section 5 states that Mistral-7B-Instruct-v0.2 and Mixtral-8x7B-Instruct-v0.1 have average invalid response rates of about 20% and 10%, respectively, in some prompt conditions, but the paper does not say whether these responses are discarded, treated as missing, or imputed, nor whether the reported percentages are computed over valid responses only. Because these models are included in the preference-rate and consequence-flipping analyses, the per-model sample sizes differ, and the comparisons in Figures 2-5 may be based on different numbers of valid responses. Please specify the exclusion rule and report per-model, per-condition sample sizes.
minor comments (6)
  1. [Abstract, Section 4.2, Section 5.4] There are several typos in model names and technical terms: 'specificed' in the abstract, 'Mistrial-7b' and 'Qwne2' in Section 4.2, and 'GPT-40' in Section 5.4. These should read 'specified', 'Mistral-7B', 'Qwen2', and 'GPT-4o' respectively.
  2. [Figures 2, 4, and 6] The figures are hard to read in the current rendering: the pair labels in Figure 2 (e.g., 'T ruth Loyalty') are ambiguous about which percentage belongs to which value, and Figure 6 has no color scale. Please add explicit value labels and legends or a color bar.
  3. [Section 3.2 and Table 6] The example in Section 3.2 and the instances in Table 4 include an 'Explanation' field, but the evaluation prompts in Table 6 do not contain this field. Please state explicitly that the Explanation is not shown to the evaluated models, so that the intended values are not leaked into the test prompts.
  4. [Section 5.2] The phrase 'scenarios where LLMs achieve unanimous agreement across all four prompts' should be stated more precisely: the unanimity is a property of a single model across the four prompt variants, not a property of the scenario independent of the model. Please clarify this in the text.
  5. [Section 4.2] Open models were run in 4-bit quantization while API models were not; this is a potential confound for cross-family and cross-size comparisons. Please discuss the likely impact of quantization, or provide a sensitivity analysis on at least one open model run at full precision.
  6. [Section 5.4] The comparison between explicit and few-shot preference handling uses different instruction wording: the explicit prompt states the preference directly, while the few-shot prompt requires the model to infer it from examples. The observed gap may therefore partly reflect instruction clarity rather than preference recognition alone. Please acknowledge this confound in the discussion.

Circularity Check

1 steps flagged · score 4.0 of 10

Moderate but partial circularity: the benchmark's value labels are LLM-generated, so the headline preference rates measure agreement with generator attributions rather than independently validated moral values.

  1. self definitional [Section 3.2 (Dataset Construction) -> Section 5.2 (Consistency, Figure 2)]
    "We specify a domain and a value pair in the prompt, and direct the model to produce diverse scenarios within the specified domain, capturing the prescribed conflicting value pair. ... for each scenario, we ask the model to generate two actions, each aligning with one value over the other."

    The action counted as 'Truth' in the 93.48% preference statistic is defined only by the generator LLM's output under the instruction 'follows on truth.' The same class of systems, including both generators (GPT-3.5 and Claude-3-Sonnet), is then asked to choose between the two generated actions, and agreement with the generator's label is reported as an LLM value preference. There is no human validation that the action actually instantiates the named value; the filtering checks only format and that the explanation mentions the two values. Therefore the headline preference rates are, by construction, rates of agreement with LLM-generated value attributions rather than an independently anchored measurement of moral values.

full rationale

This paper is an empirical evaluation rather than a mathematical derivation, so most of its pipeline is not circular: the prompt-sensitivity, consequence-flip, and explicit/few-shot alignment experiments compare model responses across controlled prompt variants and do not fit parameters or smuggle in the conclusions. The two self-citations (Singh and Singh 2023; Liscio et al. 2023) are contextual and not load-bearing. The one substantive circularity concern is the construction of the benchmark itself: the value labels that define 'Truth,' 'Loyalty,' 'Justice,' etc. are generated by LLMs and filtered only for format and keyword mention. Measuring LLM choices against these labels means the reported preferences quantify agreement between evaluator models and generator models, with the two generator models being part of the evaluated set. This is a partial, input-side circularity: the test instrument is produced by the same kind of system under test. The central claim about pronounced value preferences still has some independent empirical content, since rates differ across value pairs and across models, so the circularity is not total; however, the headline numbers should be read as model-to-model consistency rather than externally grounded moral values.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free physical parameters are fitted; this is an empirical study. The load-bearing assumptions are about the validity of the LLM-generated dilemma corpus and the interpretation of choices as value preferences.

assumptions (3)
  • domain assumption Kidder's four value pairs (Truth vs Loyalty, Individual vs Community, Short-Term vs Long-Term, Justice vs Mercy) capture the essential types of ethical dilemmas.
    The entire dataset and analysis are built on this taxonomy from Kidder (1996), which is asserted rather than empirically validated.
  • domain assumption The LLM-generated scenarios, actions, and consequences correctly instantiate the intended value conflicts and consequence valences.
    Scenarios are generated by gpt-3.5-turbo and claude-3-sonnet and filtered only by format and mention of the two values; no human validation is reported (Section 3.2).
  • domain assumption A model's choice among actions in these scenarios reflects its moral value preference rather than artifacts of the prompt wording.
    The analysis interprets action choices as value preferences, although Q1 shows substantial prompt sensitivity, so some choices may be prompt artifacts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Right vs. Right: Can LLMs Make Tough Choices?." pith.science (2026). https://pith.science/paper/ST6I3WSL

@misc{pith2026241219926,
  author       = {Pith},
  title        = {Pith review of: Right vs. Right: Can LLMs Make Tough Choices?},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ST6I3WSL}},
  note         = {Machine review of arXiv:2412.19926}
}
read the original abstract

An ethical dilemma describes a choice between two "right" options involving conflicting moral values. We present a comprehensive evaluation of how LLMs navigate ethical dilemmas. Specifically, we investigate LLMs on their (1) sensitivity in comprehending ethical dilemmas, (2) consistency in moral value choice, (3) consideration of consequences, and (4) ability to align their responses to a moral value preference explicitly or implicitly specified in a prompt. Drawing inspiration from a leading ethical framework, we construct a dataset comprising 1,730 ethical dilemmas involving four pairs of conflicting values. We evaluate 20 well-known LLMs from six families. Our experiments reveal that: (1) LLMs exhibit pronounced preferences between major value pairs, and prioritize truth over loyalty, community over individual, and long-term over short-term considerations. (2) The larger LLMs tend to support a deontological perspective, maintaining their choices of actions even when negative consequences are specified. (3) Explicit guidelines are more effective in guiding LLMs' moral choice than in-context examples. Lastly, our experiments highlight the limitation of LLMs in comprehending different formulations of ethical dilemmas.

Figures

Figures reproduced from arXiv: 2412.19926 by the authors.

Figure 1
Figure 1. Moral choice agreement for different prompts. The [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Moral value preference query results for each confl [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Percentage of flipping the choice when consequence [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Standard deviation of the percentage of choosing t [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 5
Figure 5. Figure 5: Percentage of flipping the baseline choice when the [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Heatmap of rigidity on value preferences. [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 11 canonical work pages

  1. [1]

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

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    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 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Moral foundations of large language models, 2023

    Marwa Abdulhai, Gregory Serapio-Garcia, Clément Crepy, Daria Valter, John Canny, and Natasha Jaques. Moral foundations of large language models, 2023. URL https://arxiv.org/abs/2310.15337

    arXiv 2023

  4. [4]

    Almeida, José Luiz Nunes, Neele Engelmann, Alex Wiegmann, and Marcelo de Araújo

    Guilherme F.C.F. Almeida, José Luiz Nunes, Neele Engelmann, Alex Wiegmann, and Marcelo de Araújo . Exploring the psychology of LLM s’ moral and legal reasoning. Artificial Intelligence, page 104145, 2024. ISSN 0004-3702. doi:https://doi.org/10.1016/j.artint.2024.104145. URL https://www.sciencedirect.com/science/article/pii/S000437022400081X

    arXiv 2024

  5. [5]

    Consistency and moral integrity: A self-determination theory perspective

    Alexios Arvanitis and Konstantinos Kalliris. Consistency and moral integrity: A self-determination theory perspective. Journal of Moral Education, 49 0 (3): 0 316--329, 2020. doi:10.1080/03057240.2019.1695589. URL https://doi.org/10.1080/03057240.2019.1695589

    work page arXiv 2020

  6. [6]

    Crockett, Jim Albert Charlton Everett, Theodoros Evgeniou, Alison Gopnik, Julian C

    Edmond Awad, Sydney Levine, Michael Anderson, Susan Leigh Anderson, Vincent Conitzer, Molly J. Crockett, Jim Albert Charlton Everett, Theodoros Evgeniou, Alison Gopnik, Julian C. Jamison, Tae Wan Kim, S. Matthew Liao, Michelle N. Meyer, John Mikhail, Kweku Opoku-Agyemang, Jana Schaich Borg, Juliana Schroeder, Walter Sinnott-Armstrong, Marija Slavkovik, an...

    work page 2022

  7. [7]

    Babcock, A

    William A. Babcock, A. David Gordon, and John Michael Kittross. Tools for ethical decision-making. In A. David Gordon, John Michael Kittross, John C. Merrill, William Babcock, and Michael Dorsher, editors, Controversies in Media Ethics, pages 189--208. Routledge, New York, 3 edition, 2011

    work page 2011

  8. [8]

    Using cognitive psychology to understand gpt-3

    Marcel Binz and Eric Schulz. Using cognitive psychology to understand gpt-3. Proceedings of the National Academy of Sciences, 120 0 (6): 0 e2218523120, 2023. doi:10.1073/pnas.2218523120. URL https://www.pnas.org/doi/abs/10.1073/pnas.2218523120

    doi:10.1073/pnas.2218523120 2023

Show all 35 references
  1. [9]

    Christians, Mark Fackler, Peggy J

    Clifford G. Christians, Mark Fackler, Peggy J. Kreshel, William J. Brown, Yayu Feng, Holly K. Overton, and Kathy Brittain Richardson. Media Ethics: Cases and Moral Reasoning. Routledge, New York, 12 edition, 2024

    2024

  2. [10]

    BERT : Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT : Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the Conference of the North A merican Chapter of the Association for Computational Linguistics: Human Language...

    2019 doi

  3. [11]

    Hwang, Maxwell Forbes, and Yejin Choi

    Denis Emelin, Ronan Le Bras, Jena D. Hwang, Maxwell Forbes, and Yejin Choi. Moral stories: Situated reasoning about norms, intents, actions, and their consequences. In Marie-Francine Moens, Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih, editors, Proceedings of the 2021 C...

    2021 doi

  4. [12]

    Hwang, Vered Shwartz, Maarten Sap, and Yejin Choi

    Maxwell Forbes, Jena D. Hwang, Vered Shwartz, Maarten Sap, and Yejin Choi. Social chemistry 101: Learning to reason about social and moral norms. In Bonnie Webber, Trevor Cohn, Yulan He, and Yang Liu, editors, Proceedings of the 2020 Conference on Empirical Methods in Natural ...

    2020 doi

  5. [13]

    Fraser, Svetlana Kiritchenko, and Esma Balkir

    Kathleen C. Fraser, Svetlana Kiritchenko, and Esma Balkir. Does moral code have a moral code? probing delphi ' s moral philosophy. In Apurv Verma, Yada Pruksachatkun, Kai-Wei Chang, Aram Galstyan, Jwala Dhamala, and Yang Trista Cao, editors, Proceedings of the 2nd Workshop on ...

    2022 doi

  6. [14]

    Ethical and methodological challenges in building morally informed AI systems

    Thilo Hagendorff and David Danks. Ethical and methodological challenges in building morally informed AI systems. AI and Ethics, 3: 0 1--14, 06 2022. doi:10.1007/s43681-022-00188-y

    2022 doi

  7. [15]

    Aligning AI with shared human values

    Dan Hendrycks, Collin Burns, Steven Basart, Andrew Critch, Jerry Li, Dawn Song, and Jacob Steinhardt. Aligning AI with shared human values. In International Conference on Learning Representations, 2021. URL https://openreview.net/forum?id=dNy_RKzJacY

    2021

  8. [16]

    Liwei Jiang, Jena D. Hwang, Chandra Bhagavatula, Ronan Le Bras, Jenny Liang, Jesse Dodge, Keisuke Sakaguchi, Maxwell Forbes, Jon Borchardt, Saadia Gabriel, Yulia Tsvetkov, Oren Etzioni, Maarten Sap, Regina Rini, and Yejin Choi. Can machines learn morality? T he D elphi experim...

    2022 arXiv

  9. [17]

    Rushworth M. Kidder. How Good People Make Tough Choices: Resolving the Dilemmas of Ethical Living. Simon & Schuster, New York, 1996

    1996

  10. [18]

    Enrico Liscio, Roger Lera - Leri, Filippo Bistaffa, Roel I. J. Dobbe, Catholijn M. Jonker, Maite L \' o pez - S \' a nchez, Juan A. Rodr \' guez - Aguilar, and Pradeep K. Murukannaiah. Value inference in sociotechnical systems. In Proceedings of the 22nd International Conferen...

    2023

  11. [19]

    Scruples: A corpus of community ethical judgments on 32,000 real-life anecdotes

    Nicholas Lourie, Ronan Le Bras, and Yejin Choi. Scruples: A corpus of community ethical judgments on 32,000 real-life anecdotes. Proceedings of the AAAI Conference on Artificial Intelligence, 35 0 (15): 0 13470--13479, May 2021. ISSN 2159-5399. doi:10.1609/aaai.v35i15.17589. U...

    2021 doi

  12. [20]

    Moca: measuring human-language model alignment on causal and moral judgment tasks

    Allen Nie, Yuhui Zhang, Atharva Amdekar, Chris Piech, Tatsunori Hashimoto, and Tobias Gerstenberg. Moca: measuring human-language model alignment on causal and moral judgment tasks. In Proceedings of the 37th International Conference on Neural Information Processing Systems, N...

    2024

  13. [21]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul F Christiano, Jan Leike, a...

    2022

  14. [22]

    Elicitation of situated values: Need for tools to help stakeholders and designers to reflect and communicate

    Alina Pommeranz, Christian Detweiler, Pascal Wiggers, and Catholijn Jonker. Elicitation of situated values: Need for tools to help stakeholders and designers to reflect and communicate. Ethics and Information Technology, 14 0 (4): 0 285--303, Dec 2012. ISSN 1572-8439. doi:10.1...

    2012 doi

  15. [23]

    Ethical reasoning over moral alignment: A case and framework for in-context ethical policies in LLM s

    Abhinav Rao, Aditi Khandelwal, Kumar Tanmay, Utkarsh Agarwal, and Monojit Choudhury. Ethical reasoning over moral alignment: A case and framework for in-context ethical policies in LLM s. In Houda Bouamor, Juan Pino, and Kalika Bali, editors, Findings of the Association for Co...

    2023 doi

  16. [24]

    Evaluating the moral beliefs encoded in LLM s

    Nino Scherrer, Claudia Shi, Amir Feder, and David Blei. Evaluating the moral beliefs encoded in LLM s. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems, volume 36, pages 51778--51809. Curran Asso...

    2023

  17. [25]

    BERT has a moral compass: I mprovements of ethical and moral values of machines, 2019

    Patrick Schramowski, Cigdem Turan, Sophie Jentzsch, Constantin Rothkopf, and Kristian Kersting. BERT has a moral compass: I mprovements of ethical and moral values of machines, 2019. URL https://arxiv.org/abs/1912.05238

    2019 arXiv

  18. [26]

    Quantifying language models' sensitivity to spurious features in prompt design or: How i learned to start worrying about prompt formatting

    Melanie Sclar, Yejin Choi, Yulia Tsvetkov, and Alane Suhr. Quantifying language models' sensitivity to spurious features in prompt design or: How i learned to start worrying about prompt formatting. In The Twelfth International Conference on Learning Representations, 2024. URL...

    2024

  19. [27]

    Singh and Munindar P

    Amika M. Singh and Munindar P. Singh. Wasabi: A conceptual model for trustworthy artificial intelligence. IEEE Computer, 56 0 (2): 0 20--28, February 2023. doi:10.1109/MC.2022.3212022

    2023

  20. [28]

    Hwang, Sydney Levine, Valentina Pyatkin, Peter West, Nouha Dziri, Ximing Lu, Kavel Rao, Chandra Bhagavatula, Maarten Sap, John Tasioulas, and Yejin Choi

    Taylor Sorensen, Liwei Jiang, Jena D. Hwang, Sydney Levine, Valentina Pyatkin, Peter West, Nouha Dziri, Ximing Lu, Kavel Rao, Chandra Bhagavatula, Maarten Sap, John Tasioulas, and Yejin Choi. Value kaleidoscope: Engaging ai with pluralistic human values, rights, and duties. Pr...

    2024 doi

  21. [29]

    On the machine learning of ethical judgments from natural language

    Zeerak Talat, Hagen Blix, Josef Valvoda, Maya Indira Ganesh, Ryan Cotterell, and Adina Williams. On the machine learning of ethical judgments from natural language. In Marine Carpuat, Marie-Catherine de Marneffe, and Ivan Vladimir Meza Ruiz, editors, Proceedings of the 2022 Co...

    2022

  22. [30]

    Probing the moral development of large language models through defining issues test, 2023

    Kumar Tanmay, Aditi Khandelwal, Utkarsh Agarwal, and Monojit Choudhury. Probing the moral development of large language models through defining issues test, 2023. URL https://arxiv.org/abs/2309.13356

    2023 arXiv

  23. [31]

    Would you rather? a new benchmark for learning machine alignment with cultural values and social preferences

    Yi Tay, Donovan Ong, Jie Fu, Alvin Chan, Nancy Chen, Anh Tuan Luu, and Chris Pal. Would you rather? a new benchmark for learning machine alignment with cultural values and social preferences. In Dan Jurafsky, Joyce Chai, Natalie Schluter, and Joel Tetreault, editors, Proceedin...

    2020 doi

  24. [32]

    Large language models are not fair evaluators

    Peiyi Wang, Lei Li, Liang Chen, Dawei Zhu, Binghuai Lin, Yunbo Cao, Qi Liu, Tianyu Liu, and Zhifang Sui. Large language models are not fair evaluators. ArXiv, abs/2305.17926, 2023. URL https://api.semanticscholar.org/CorpusID:258960339

    2023 arXiv

  25. [33]

    Dai, and Quoc V Le

    Jason Wei, Maarten Bosma, Vincent Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai, and Quoc V Le. Finetuned language models are zero-shot learners. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=gEZrGCozdqR

    2022

  26. [34]

    Jingyan Zhou, Minda Hu, Junan Li, Xiaoying Zhang, Xixin Wu, Irwin King, and Helen M. Meng. Rethinking machine ethics - C an LLM s perform moral reasoning through the lens of moral theories? ArXiv, abs/2308.15399, 2023. URL https://api.semanticscholar.org/CorpusID:261276143

    2023 arXiv

  27. [35]

    N orm B ank: A knowledge bank of situational social norms

    Caleb Ziems, Jane Dwivedi-Yu, Yi-Chia Wang, Alon Halevy, and Diyi Yang. N orm B ank: A knowledge bank of situational social norms. In Anna Rogers, Jordan Boyd-Graber, and Naoaki Okazaki, editors, Proceedings of the 61st Annual Meeting of the Association for Computational Lingu...

    2023 doi

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.