Moral Mimicry: Large Language Models Produce Moral Rationalizations Tailored to Political Identity

Simmons, Gabriel

doi:10.18653/v1/2023.acl-srw.40

Cited by 9 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Researchers have introduced harm taxonomies specifically for LLMs, which identify known risks (i.e., informed by observed instances of harm) [18,100,190] and emerging risks of LLMs (anticipated risks based on foreseeable capabilities of LLMs) [108,166]. Since LLMs can be used for a wide range of tasks associated with many different categories of harms, researchers have presented frameworks and evaluation methods to assess a particular type of LLM harm, including misinformation [74,135], representation and toxicity [42,64], human autonomy [65,168], malicious use [38,154], and data privacy [87,97]. The popular methods to identify these harms include benchmarking [27,28], user research [101,106], and adversarial testing [41,137].…”

Section: Identifying and Mitigating Llm Harmsmentioning

confidence: 99%

Farsight: Fostering Responsible AI Awareness During AI Application Prototyping

Wang,

Kulkarni,

Wilcox

et al. 2024

Proceedings of the CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

Fig. 1: With in situ interfaces and novel techniques, Farsight empowers AI prototypers to envision potential harms that may arise from their large language models (LLMs)-powered AI applications during early prototyping. (A) In this example, an AI prototyper is creating a prompt for an English-to-French translator in a web-based AI prototyping tool. (B) The Alert Symbol from Farsight warns the user of potential risks associated with their AI application. (C) Clicking the symbol expands the Awareness Sidebar, highlighting news articles relevant to the user's prompt (top), and LLM-generated potential use cases and harms (bottom). (D) Clicking the blue button opens the Harm Envisioner that allows the user to interactively envision, assess, and reflect on the potential use cases, stakeholders, and harms of their AI application with the assistance of an LLM.

show abstract

Section: Identifying and Mitigating Llm Harmsmentioning

confidence: 99%

Farsight: Fostering Responsible AI Awareness During AI Application Prototyping

Wang,

Kulkarni,

Wilcox

et al. 2024

Proceedings of the CHI Conference on Human Factors in Computing Systems

View full text Add to dashboard Cite

show abstract

“…Foundation models in particular can increase the scale and speed at which disinformation campaigns can be disseminated across the information ecosystem [161,198]. As generative AI applications powered by foundation models flood the public sphere with fake information, there is a risk of eroding public trust in the information that circulates online, further fueling social polarization and the creation of echo chambers [113,195].…”

Section: Social Risks and Harmsmentioning

confidence: 99%

Mapping the individual, social and biospheric impacts of Foundation Models

Domínguez Hernández,

Krishna,

Perini

et al. 2024

The 2024 ACM Conference on Fairness, Accountability, and Transparency

View full text Add to dashboard Cite

Responding to the rapid roll-out and large-scale commercialization of foundation models, large language models, and generative AI, an emerging body of work is shedding light on the myriad impacts these technologies are having across society. Such research is expansive, ranging from the production of discriminatory, fake and toxic outputs, and privacy and copyright violations, to the unjust extraction of labor and natural resources. The same has not been the case in some of the most prominent AI governance initiatives in the global north like the UK's AI Safety Summit and the G7's Hiroshima process, which have influenced much of the international dialogue around AI governance. Despite the wealth of cautionary tales and evidence of algorithmic harm, there has been an ongoing over-emphasis within the AI governance discourse on technical matters of safety and global catastrophic or existential risks. This narrowed focus has tended to draw attention away from very pressing social and ethical challenges posed by the current brute-force industrialization of AI applications. To address such a visibility gap between real-world consequences and speculative risks, this paper offers a critical framework to account for the social, political, and environmental dimensions of foundation models and generative AI. Drawing on a review of the literature on the harms and risks of * Equal contribution as lead authors † Also with The Alan Turing Institute.

show abstract

“…Given the increasing societal role played by Large Language Models (LLMs), researchers have begun to investigate the underlying psychology of these generative models. For example, several works have investigated whether LLMs can truly understand language and perform reasoning (Chowdhery et al, 2022), understand distinctions between different moralities and personalities (Miotto et al, 2022;Simmons, 2022), and learn ethical dilemmas (Jiang et al, 2021). Hagendorff et al (2022), for instance, demonstrated that LLMs are intuitive decision makers, just like humans, arguing that investigating LLMs with methods from psychology has the potential to uncover their emergent traits and behavior.…”

Section: Introductionmentioning

confidence: 99%

Which Humans?

Atari,

Xue,

Park

et al. 2023

Preprint

View full text Add to dashboard Cite

Large language models (LLMs) have recently made vast advances in both generating and analyzing textual data. Technical reports often compare LLMs’ outputs with “human” performance on various tests. Here, we ask, “Which humans?” Much of the existing literature largely ignores the fact that humans are a cultural species with substantial psychological diversity around the globe that is not fully captured by the textual data on which current LLMs have been trained. We show that LLMs’ responses to psychological measures are an outlier compared with large-scale cross-cultural data, and that their performance on cognitive psychological tasks most resembles that of people from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies but declines rapidly as we move away from these populations (r = -.70). Ignoring cross-cultural diversity in both human and machine psychology raises numerous scientific and ethical issues. We close by discussing ways to mitigate the WEIRD bias in future generations of generative language models.

show abstract

Moral Mimicry: Large Language Models Produce Moral Rationalizations Tailored to Political Identity

Cited by 9 publications

References 0 publications

Farsight: Fostering Responsible AI Awareness During AI Application Prototyping

Farsight: Fostering Responsible AI Awareness During AI Application Prototyping

Mapping the individual, social and biospheric impacts of Foundation Models

Which Humans?

Contact Info

Product

Resources

About