Explicit and Implicit Large Language Model Personas Generate Opinions but Fail to Replicate Deeper Perceptions and Biases
2406.14462
0
0
Abstract
Large language models (LLMs) are increasingly being used in human-centered social scientific tasks, such as data annotation, synthetic data creation, and engaging in dialog. However, these tasks are highly subjective and dependent on human factors, such as one's environment, attitudes, beliefs, and lived experiences. Thus, employing LLMs (which do not have such human factors) in these tasks may result in a lack of variation in data, failing to reflect the diversity of human experiences. In this paper, we examine the role of prompting LLMs with human-like personas and asking the models to answer as if they were a specific human. This is done explicitly, with exact demographics, political beliefs, and lived experiences, or implicitly via names prevalent in specific populations. The LLM personas are then evaluated via (1) subjective annotation task (e.g., detecting toxicity) and (2) a belief generation task, where both tasks are known to vary across human factors. We examine the impact of explicit vs. implicit personas and investigate which human factors LLMs recognize and respond to. Results show that LLM personas show mixed results when reproducing known human biases, but generate generally fail to demonstrate implicit biases. We conclude that LLMs lack the intrinsic cognitive mechanisms of human thought, while capturing the statistical patterns of how people speak, which may restrict their effectiveness in complex social science applications.
Create account to get full access
Overview
- This paper investigates the ability of large language models (LLMs) to generate opinions and replicate deeper perceptions and biases when given explicit or implicit personas.
- The researchers find that LLMs can generate opinions that align with given personas, but struggle to fully capture the nuanced biases and perceptions that would be expected from a human with that persona.
- The paper provides insights into the limitations of current LLM technology in accurately simulating human-like social and cognitive behaviors.
Plain English Explanation
In this study, the researchers explored how well large language models (LLMs) - powerful AI systems trained on massive amounts of text data - can generate opinions and beliefs that align with specific personas or personality profiles. The researchers gave the LLMs explicit instructions to adopt certain personas, such as being a conservative or liberal political supporter, and then analyzed the opinions and viewpoints the models produced.
The key finding is that the LLMs were able to generate opinions that were generally consistent with the assigned personas. For example, when given a "conservative" persona, the model would express views typical of political conservatism. However, the researchers found that the LLMs struggled to fully capture the deeper perceptions, biases, and nuanced thought processes that would be expected from a human with that persona.
In other words, the LLMs could mimic the surface-level opinions and language associated with a persona, but they failed to replicate the more complex cognitive and social behaviors that would characterize a human with that persona. This suggests that current LLM technology, while impressive in many ways, still has limitations in truly understanding and embodying human-like perspectives and decision-making.
The implications of this research are important as LLMs become more prominent in applications like customer service, content generation, and even political discourse. It highlights the need for continued advancements in AI systems to better capture the richness and complexity of human cognition and behavior.
Technical Explanation
The researchers conducted a series of experiments to evaluate the ability of LLMs to generate opinions and beliefs that align with explicit and implicit personas. They used two popular LLM architectures, GPT-3 and BERT, and provided them with prompts instructing the models to adopt specific personas, such as a conservative or liberal political supporter.
The researchers then analyzed the opinions and viewpoints expressed by the LLMs in response to a variety of prompts and questions. They found that the LLMs were able to generate responses that were generally consistent with the assigned personas, demonstrating an ability to align their outputs with explicit persona cues.
However, the researchers also found that the LLMs struggled to fully capture the deeper perceptions, biases, and nuanced thought processes that would be expected from a human with a particular persona. This suggests that current LLMs, while capable of generating persona-aligned opinions, lack the ability to truly replicate the complex social and cognitive behaviors that characterize human-like personas.
The findings from this study contribute to a growing body of research exploring the limitations of LLMs in simulating human-like social and cognitive behaviors. This includes previous work on quantifying the persona effect in LLM simulations and evaluating the biases present in LLM outputs.
Critical Analysis
The researchers acknowledge several caveats and limitations in their study. Firstly, they note that the persona cues provided to the LLMs may have been too explicit, and that more subtle or implicit persona signals may be required to better capture the nuanced cognitive and social behaviors associated with human-like personas.
Additionally, the researchers suggest that the LLMs may have been limited by the training data they were exposed to, which may not have adequately represented the full range of human perspectives and biases. Further research could explore the impact of different training datasets and techniques on the LLMs' ability to simulate human-like personas.
It's also important to consider the potential ethical implications of LLMs generating persona-aligned opinions, especially in sensitive domains such as politics or social issues. While the researchers did not explore these implications in depth, the findings highlight the need for careful consideration of how these systems are deployed and the potential for misuse or manipulation.
Overall, this study provides valuable insights into the current limitations of LLMs in replicating the depth and complexity of human cognition and behavior. As these systems continue to advance, it will be crucial to address these limitations and ensure that they are developed and deployed in a responsible and ethical manner.
Conclusion
This paper offers important insights into the capabilities and limitations of large language models (LLMs) in generating opinions and beliefs that align with specific personas. The researchers found that while LLMs can produce responses that are generally consistent with assigned personas, they struggle to fully capture the deeper perceptions, biases, and nuanced thought processes that would be expected from a human with that persona.
These findings contribute to a growing body of research exploring the challenges of simulating human-like social and cognitive behaviors with LLMs. As these powerful AI systems become more prevalent in various applications, understanding their limitations is crucial to ensure they are developed and deployed responsibly and ethically.
The paper serves as a reminder that while LLMs have made impressive strides in natural language processing, there is still significant work to be done to truly replicate the richness and complexity of human cognition and behavior. Continued advancements in AI research, particularly in the areas of social and cognitive modeling, will be essential in addressing these limitations and unlocking the full potential of large language models.
This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!
Related Papers
Evaluating Large Language Model Biases in Persona-Steered Generation
Andy Liu, Mona Diab, Daniel Fried
0
0
The task of persona-steered text generation requires large language models (LLMs) to generate text that reflects the distribution of views that an individual fitting a persona could have. People have multifaceted personas, but prior work on bias in LLM-generated opinions has only explored multiple-choice settings or one-dimensional personas. We define an incongruous persona as a persona with multiple traits where one trait makes its other traits less likely in human survey data, e.g. political liberals who support increased military spending. We find that LLMs are 9.7% less steerable towards incongruous personas than congruous ones, sometimes generating the stereotypical stance associated with its demographic rather than the target stance. Models that we evaluate that are fine-tuned with Reinforcement Learning from Human Feedback (RLHF) are more steerable, especially towards stances associated with political liberals and women, but present significantly less diverse views of personas. We also find variance in LLM steerability that cannot be predicted from multiple-choice opinion evaluation. Our results show the importance of evaluating models in open-ended text generation, as it can surface new LLM opinion biases. Moreover, such a setup can shed light on our ability to steer models toward a richer and more diverse range of viewpoints.
5/31/2024
Are Large Language Models Chameleons?
Mingmeng Geng, Sihong He, Roberto Trotta
0
0
Do large language models (LLMs) have their own worldviews and personality tendencies? Simulations in which an LLM was asked to answer subjective questions were conducted more than 1 million times. Comparison of the responses from different LLMs with real data from the European Social Survey (ESS) suggests that the effect of prompts on bias and variability is fundamental, highlighting major cultural, age, and gender biases. Methods for measuring the difference between LLMs and survey data are discussed, such as calculating weighted means and a new proposed measure inspired by Jaccard similarity. We conclude that it is important to analyze the robustness and variability of prompts before using LLMs to model individual decisions or collective behavior, as their imitation abilities are approximate at best.
5/30/2024
💬
PersonaLLM: Investigating the Ability of Large Language Models to Express Personality Traits
Hang Jiang, Xiajie Zhang, Xubo Cao, Cynthia Breazeal, Deb Roy, Jad Kabbara
0
0
Despite the many use cases for large language models (LLMs) in creating personalized chatbots, there has been limited research on evaluating the extent to which the behaviors of personalized LLMs accurately and consistently reflect specific personality traits. We consider studying the behavior of LLM-based agents which we refer to as LLM personas and present a case study with GPT-3.5 and GPT-4 to investigate whether LLMs can generate content that aligns with their assigned personality profiles. To this end, we simulate distinct LLM personas based on the Big Five personality model, have them complete the 44-item Big Five Inventory (BFI) personality test and a story writing task, and then assess their essays with automatic and human evaluations. Results show that LLM personas' self-reported BFI scores are consistent with their designated personality types, with large effect sizes observed across five traits. Additionally, LLM personas' writings have emerging representative linguistic patterns for personality traits when compared with a human writing corpus. Furthermore, human evaluation shows that humans can perceive some personality traits with an accuracy of up to 80%. Interestingly, the accuracy drops significantly when the annotators were informed of AI authorship.
4/3/2024
Large Language Models Show Human-like Social Desirability Biases in Survey Responses
Aadesh Salecha, Molly E. Ireland, Shashanka Subrahmanya, Jo~ao Sedoc, Lyle H. Ungar, Johannes C. Eichstaedt
0
0
As Large Language Models (LLMs) become widely used to model and simulate human behavior, understanding their biases becomes critical. We developed an experimental framework using Big Five personality surveys and uncovered a previously undetected social desirability bias in a wide range of LLMs. By systematically varying the number of questions LLMs were exposed to, we demonstrate their ability to infer when they are being evaluated. When personality evaluation is inferred, LLMs skew their scores towards the desirable ends of trait dimensions (i.e., increased extraversion, decreased neuroticism, etc). This bias exists in all tested models, including GPT-4/3.5, Claude 3, Llama 3, and PaLM-2. Bias levels appear to increase in more recent models, with GPT-4's survey responses changing by 1.20 (human) standard deviations and Llama 3's by 0.98 standard deviations-very large effects. This bias is robust to randomization of question order and paraphrasing. Reverse-coding all the questions decreases bias levels but does not eliminate them, suggesting that this effect cannot be attributed to acquiescence bias. Our findings reveal an emergent social desirability bias and suggest constraints on profiling LLMs with psychometric tests and on using LLMs as proxies for human participants.
5/13/2024