Search for: All records

The value alignment of sociotechnical systems has become a central debate, but progress depends on how human values are perceived in the content these systems surface and how such perceptions can be measured at scale. Social media platforms are a prominent class of sociotechnical systems where algorithmic curation shapes exposure to value-laden content at scale. Large-language models offer new opportunities for measuring expressions of human values (e.g., humility or equality) in social media data, but value expressions can be subjective: different people will annotate the same post with different values. In this paper, we draw on the Schwartz value system as a broadly encompassing and theoretically grounded set of basic human values, and introduce a framework to personalize the measurement of expressions of Schwartz values in social media posts at scale. We collect 32,370 ground truth value expression annotations from N=1,079 people on 5,211 social media posts representative of real users' feeds. Due to the subjectivity of the task, we observe low levels of inter-rater agreement between people, and low agreement between human raters and LLM-based methods. In response, we construct a personalization architecture for classifying value expressions by learning from a small number of highly informative calibration annotations per user. In evaluation, we find that modeling these differences successfully yields value expression predictions that people agree with more than they agree with other people. These results contribute new methods and understanding for the measurement of human values in social media data. 

Users trust algorithms more when they can predict the algorithms’ behavior. Simple algorithms trivially yield predictively accurate mental models, but modern AI algorithms have often been assumed too complex for people to build predictive mental models, especially in the social media domain. In this paper, we describe conditions under which even complex algorithms can yield predictive mental models, opening up opportunities for a broader set of human-centered algorithms. We theorize that users will form an accurate predictive mental model of an algorithm’s behavior if and only if the algorithm simultaneously satisfies three criteria: (1) cognitive availability of the underlying concepts being modeled, (2) concept compactness (does it form a single cognitive construct?), and (3) high alignment between the person’s and algorithm’s execution of the concept. We evaluate this theory through a pre-registered experiment (N= 1250) where users predict behavior of 25 social media feed ranking algorithms that vary on these criteria. We find that even complex (e.g., LLM-based) algorithms enjoy accurate prediction rates when they meet all criteria, and even simple (e.g., basic term count) algorithms fail to be predictable when a single criterion fails. We also find that these criteria determine outcomes beyond prediction accuracy, such as which mental models users deploy to make their predictions. 

Large language models promise a broad set of functions, but when not given a specific objective, they default to generic results. We demonstrate that inferring the user’s in-the-moment objective, then rapidly optimizing for that singular objective, enables LLMs to produce specialized tools, interfaces, and responses. Our work introduces just-in-time objectives, which model a user’s goals to specialize LLM systems on the fly. We contribute an architecture for automatically inducing such objectives by passively observing user behavior, then steering downstream AI systems through generation and evaluation against this objective. Inducing just-in-time objectives (e.g., “Clarify the abstract’s research contribution”) enables automatic generation of tools, e.g., those that critique a draft based on relevant HCI methodologies, anticipate related researchers’ reactions, or surface ambiguous terminology. In a series of experiments on participants’ own tasks, JIT objectives enable LLM outputs that achieve 66–86% win rates over typical LLMs. In-person use sessions confirm that JIT objectives produce specialized tools that are unique to each participant and are rated as significantly higher quality than a standard LLM chat tool. 

Online marketplaces use rating systems to promote the discovery of high-quality products. However, these systems also lead to high variance in producers' economic outcomes: a new producer who sells high-quality items, may unluckily receive a low rating early, severely impacting their future popularity. We investigate the design of rating systems that balance the goals of identifying high-quality products (``efficiency'') and minimizing the variance in outcomes of producers of similar quality (individual ``producer fairness'').We show that there is a trade-off between these two goals: rating systems that promote efficiency are necessarily less individually fair to producers. We introduce prior-weighted rating systems as an approach to managing this trade-off. Informally, the system we propose sets a system-wide prior for the quality of an incoming product; subsequently, the system updates that prior to a posterior for each product's quality based on user-generated ratings over time. We show theoretically that in markets where products accrue reviews at an equal rate, the strength of the rating system's prior determines the operating point on the identified trade-off: the stronger the prior, the more the marketplace discounts early ratings data (increasing individual fairness), but the slower the platform is in learning about true item quality (so efficiency suffers). We further analyze this trade-off in a responsive market where customers make decisions based on historical ratings. Through calibrated simulations in 19 different real-world datasets sourced from large online platforms, we show that the choice of prior strength mediates the same efficiency-consistency trade-off in this setting. Overall, we demonstrate that by tuning the prior as a design choice in a prior-weighted rating system, platforms can be intentional about the balance between efficiency and producer fairness. 

Today, social media platforms hold the sole power to study the effects of feed-ranking algorithms. We developed a platform-independent method that reranks participants’ feeds in real time and used this method to conduct a preregistered 10-day field experiment with 1256 participants on X during the 2024 US presidential campaign. Our experiment used a large language model to rerank posts that expressed antidemocratic attitudes and partisan animosity (AAPA). Decreasing or increasing AAPA exposure shifted out-party partisan animosity by more than 2 points on a 100-point feeling thermometer, with no detectable differences across party lines, providing causal evidence that exposure to AAPA content alters affective polarization. This work establishes a method to study feed algorithms without requiring platform cooperation, enabling independent evaluation of ranking interventions in naturalistic settings. 

Social media feed ranking algorithms fail when they too narrowly focus on engagement as their objective. The literature has asserted a wide variety of values that these algorithms should account for as well -- ranging from well-being to productive discourse -- far more than can be encapsulated by a single topic or theory. In response, we present a library of values for social media algorithms: a pluralistic set of 78 values as articulated across the literature, implemented into LLM-powered content classifiers that can be installed individually or in combination for real-time re-ranking of social media feeds. We investigate this approach by developing a browser extension, Alexandria, that re-ranks the X/Twitter feed in real time based on the user's desired values. Through two user studies, both qualitative (N=12) and quantitative (N=257), we found that diverse user needs require a large library of values, enabling more nuanced preferences and greater user control. With this work, we argue that the values criticized as missing from social media ranking algorithms can be operationalized and deployed today through end-user tools. 

Language models are aligned to emulate the collective voice of many, resulting in outputs that align with no one in particular. Steering LLMs away from generic output is possible through supervised finetuning or RLHF, but requires prohibitively large datasets for new ad-hoc tasks. We argue that it is instead possible to align an LLM to a specific setting by leveraging a very small number (< 10) of demonstrations as feedback. Our method, Demonstration ITerated Task Optimization (DITTO), directly aligns language model outputs to a user's demonstrated behaviors. Derived using ideas from online imitation learning, DITTO cheaply generates online comparison data by treating users' demonstrations as preferred over output from the LLM and its intermediate checkpoints. Concretely, DITTO operates by having an LLM generate examples that are presumed to be inferior to expert demonstrations. The method iteratively constructs pairwise preference relationships between these LLM-generated samples and expert demonstrations, potentially including comparisons between different training checkpoints. These constructed preference pairs are then used to train the model using a preference optimization algorithm (e.g. DPO). We evaluate DITTO's ability to learn fine-grained style and task alignment across domains such as news articles, emails, and blog posts. Additionally, we conduct a user study soliciting a range of demonstrations from participants (N = 16). Across our benchmarks and user study, we find that win-rates for DITTO outperform few-shot prompting, supervised fine-tuning, and other self-play methods by an avg. of 19% points. By using demonstrations as feedback directly, DITTO offers a novel method for effective customization of LLMs.