Mitchell Gordon

Jillian Fisher, Skyler Hallinan, Ximing Lu et al. · uw

Authorship obfuscation, rewriting a text to intentionally obscure the identity of the author, is an important but challenging task. Current methods using large language models (LLMs) lack interpretability and controllability, often ignoring author-specific stylistic features, resulting in less robust performance overall. To address this, we develop StyleRemix, an adaptive and interpretable obfuscation method that perturbs specific, fine-grained style elements of the original input text. StyleRemix uses pre-trained Low Rank Adaptation (LoRA) modules to rewrite an input specifically along various stylistic axes (e.g., formality and length) while maintaining low computational cost. StyleRemix outperforms state-of-the-art baselines and much larger LLMs in a variety of domains as assessed by both automatic and human evaluation. Additionally, we release AuthorMix, a large set of 30K high-quality, long-form texts from a diverse set of 14 authors and 4 domains, and DiSC, a parallel corpus of 1,500 texts spanning seven style axes in 16 unique directions

Andre Ye, Jenny Y. Huang, Alicia Guo et al.

When language models answer open-ended problems, they implicitly make hidden decisions that shape their outputs, leaving users with uncontextualized answers rather than a working map of the problem; drawing on multiverse analysis from statistics, we build and evaluate the conceptual multiverse, an interactive system that represents conceptual decisions such as how to frame a question or what to value as a space users can transparently inspect, intervenably change, and check against principled domain reasoning; for this structure to be worth navigating rather than misleading, it must be rigorous and checkable against domain reasoning norms, so we develop a general verification framework that enforces properties of good decision structures like unambiguity and completeness calibrated by expert-level reasoning; across three domains, the conceptual multiverse helped participants develop a working map of the problem, with philosophy students rewriting essays with sharper framings and reversed theses, alignment annotators moving from surface preferences to reasoning about user intent and harm, and poets identifying compositional patterns that clarified their taste.

Taylor Sorensen, Jared Moore, Jillian Fisher et al. · uw

With increased power and prevalence of AI systems, it is ever more critical that AI systems are designed to serve all, i.e., people with diverse values and perspectives. However, aligning models to serve pluralistic human values remains an open research question. In this piece, we propose a roadmap to pluralistic alignment, specifically using language models as a test bed. We identify and formalize three possible ways to define and operationalize pluralism in AI systems: 1) Overton pluralistic models that present a spectrum of reasonable responses; 2) Steerably pluralistic models that can steer to reflect certain perspectives; and 3) Distributionally pluralistic models that are well-calibrated to a given population in distribution. We also formalize and discuss three possible classes of pluralistic benchmarks: 1) Multi-objective benchmarks, 2) Trade-off steerable benchmarks, which incentivize models to steer to arbitrary trade-offs, and 3) Jury-pluralistic benchmarks which explicitly model diverse human ratings. We use this framework to argue that current alignment techniques may be fundamentally limited for pluralistic AI; indeed, we highlight empirical evidence, both from our own experiments and from other work, that standard alignment procedures might reduce distributional pluralism in models, motivating the need for further research on pluralistic alignment.

Y. Alex Kolchinski, Sharon Zhou, Shengjia Zhao et al.

Generative models have made immense progress in recent years, particularly in their ability to generate high quality images. However, that quality has been difficult to evaluate rigorously, with evaluation dominated by heuristic approaches that do not correlate well with human judgment, such as the Inception Score and Fréchet Inception Distance. Real human labels have also been used in evaluation, but are inefficient and expensive to collect for each image. Here, we present a novel method to automatically evaluate images based on their quality as perceived by humans. By not only generating image embeddings from Inception network activations and comparing them to the activations for real images, of which other methods perform a variant, but also regressing the activation statistics to match gold standard human labels, we demonstrate 66% accuracy in predicting human scores of image realism, matching the human inter-rater agreement rate. Our approach also generalizes across generative models, suggesting the potential for capturing a model-agnostic measure of image quality. We open source our dataset of human labels for the advancement of research and techniques in this area.

Niklas Stoehr, Mitchell Gordon, Chiyuan Zhang et al. · eth-zurich

Can we localize the weights and mechanisms used by a language model to memorize and recite entire paragraphs of its training data? In this paper, we show that while memorization is spread across multiple layers and model components, gradients of memorized paragraphs have a distinguishable spatial pattern, being larger in lower model layers than gradients of non-memorized examples. Moreover, the memorized examples can be unlearned by fine-tuning only the high-gradient weights. We localize a low-layer attention head that appears to be especially involved in paragraph memorization. This head is predominantly focusing its attention on distinctive, rare tokens that are least frequent in a corpus-level unigram distribution. Next, we study how localized memorization is across the tokens in the prefix by perturbing tokens and measuring the caused change in the decoding. A few distinctive tokens early in a prefix can often corrupt the entire continuation. Overall, memorized continuations are not only harder to unlearn, but also to corrupt than non-memorized ones.

Yu Ying Chiu, Michael S. Lee, Rachel Calcott et al. · uw

As AI systems progress, we rely more on them to make decisions with us and for us. To ensure that such decisions are aligned with human values, it is imperative for us to understand not only what decisions they make but also how they come to those decisions. Reasoning language models, which provide both final responses and (partially transparent) intermediate thinking traces, present a timely opportunity to study AI procedural reasoning. Unlike math and code problems which often have objectively correct answers, moral dilemmas are an excellent testbed for process-focused evaluation because they allow for multiple defensible conclusions. To do so, we present MoReBench: 1,000 moral scenarios, each paired with a set of rubric criteria that experts consider essential to include (or avoid) when reasoning about the scenarios. MoReBench contains over 23 thousand criteria including identifying moral considerations, weighing trade-offs, and giving actionable recommendations to cover cases on AI advising humans moral decisions as well as making moral decisions autonomously. Separately, we curate MoReBench-Theory: 150 examples to test whether AI can reason under five major frameworks in normative ethics. Our results show that scaling laws and existing benchmarks on math, code, and scientific reasoning tasks fail to predict models' abilities to perform moral reasoning. Models also show partiality towards specific moral frameworks (e.g., Benthamite Act Utilitarianism and Kantian Deontology), which might be side effects of popular training paradigms. Together, these benchmarks advance process-focused reasoning evaluation towards safer and more transparent AI.

Mitchell L. Gordon, Michelle S. Lam, Joon Sung Park et al.

Whose labels should a machine learning (ML) algorithm learn to emulate? For ML tasks ranging from online comment toxicity to misinformation detection to medical diagnosis, different groups in society may have irreconcilable disagreements about ground truth labels. Supervised ML today resolves these label disagreements implicitly using majority vote, which overrides minority groups' labels. We introduce jury learning, a supervised ML approach that resolves these disagreements explicitly through the metaphor of a jury: defining which people or groups, in what proportion, determine the classifier's prediction. For example, a jury learning model for online toxicity might centrally feature women and Black jurors, who are commonly targets of online harassment. To enable jury learning, we contribute a deep learning architecture that models every annotator in a dataset, samples from annotators' models to populate the jury, then runs inference to classify. Our architecture enables juries that dynamically adapt their composition, explore counterfactuals, and visualize dissent.

Sharon Zhou, Mitchell L. Gordon, Ranjay Krishna et al.

Generative models often use human evaluations to measure the perceived quality of their outputs. Automated metrics are noisy indirect proxies, because they rely on heuristics or pretrained embeddings. However, up until now, direct human evaluation strategies have been ad-hoc, neither standardized nor validated. Our work establishes a gold standard human benchmark for generative realism. We construct Human eYe Perceptual Evaluation (HYPE) a human benchmark that is (1) grounded in psychophysics research in perception, (2) reliable across different sets of randomly sampled outputs from a model, (3) able to produce separable model performances, and (4) efficient in cost and time. We introduce two variants: one that measures visual perception under adaptive time constraints to determine the threshold at which a model's outputs appear real (e.g. 250ms), and the other a less expensive variant that measures human error rate on fake and real images sans time constraints. We test HYPE across six state-of-the-art generative adversarial networks and two sampling techniques on conditional and unconditional image generation using four datasets: CelebA, FFHQ, CIFAR-10, and ImageNet. We find that HYPE can track model improvements across training epochs, and we confirm via bootstrap sampling that HYPE rankings are consistent and replicable.