As foundation models have accumulated hundreds of millions of users, developers have begun to take steps to prevent harmful types of uses. One salient intervention that foundation model developers adopt is acceptable use policies: legally binding policies that prohibit users from using a model for specific purposes. This paper identifies acceptable use policies from 30 foundation model developers, analyzes the use restrictions they contain, and argues that acceptable use policies are an important lens for understanding the regulation of foundation models. Taken together, developers' acceptable use policies include 127 distinct use restrictions; the wide variety in the number and type of use restrictions may create fragmentation across the AI supply chain. Developers also employ acceptable use policies to prevent competitors or specific industries from making use of their models. Developers alone decide what constitutes acceptable use, and rarely provide transparency about how they enforce their policies. In practice, acceptable use policies are difficult to enforce, and scrupulous enforcement can act as a barrier to researcher access and limit beneficial uses of foundation models. Nevertheless, acceptable use policies for foundation models are an early example of self-regulation that have a significant impact on the market for foundation models and the overall AI ecosystem.

Large language models (LLMs) are perceived by some as having the potential to revolutionize social science research, considering their training data includes information on human attitudes and behavior. If these attitudes are reflected in LLM output, LLM-generated "synthetic samples" could be used as a viable and efficient alternative to surveys of real humans. However, LLM-synthetic samples might exhibit coverage bias due to training data and fine-tuning processes being unrepresentative of diverse linguistic, social, political, and digital contexts. In this study, we examine to what extent LLM-based predictions of public opinion exhibit context-dependent biases by predicting voting behavior in the 2024 European Parliament elections using a state-of-the-art LLM. We prompt GPT-4-Turbo with anonymized individual-level background information, varying prompt content and language, ask the LLM to predict each person's voting behavior, and compare the weighted aggregates to the real election results. Our findings emphasize the limited applicability of LLM-synthetic samples to public opinion prediction. We show that (1) the LLM-based prediction of future voting behavior largely fails, (2) prediction accuracy is unequally distributed across national and linguistic contexts, and (3) improving LLM predictions requires detailed attitudinal information about individuals for prompting. In investigating the contextual differences of LLM-based predictions of public opinion, our research contributes to the understanding and mitigation of biases and inequalities in the development of LLMs and their applications in computational social science.

The dominant practice of AI alignment assumes (1) that preferences are an adequate representation of human values, (2) that human rationality can be understood in terms of maximizing the satisfaction of preferences, and (3) that AI systems should be aligned with the preferences of one or more humans to ensure that they behave safely and in accordance with our values. Whether implicitly followed or explicitly endorsed, these commitments constitute what we term a preferentist approach to AI alignment. In this paper, we characterize and challenge the preferentist approach, describing conceptual and technical alternatives that are ripe for further research. We first survey the limits of rational choice theory as a descriptive model, explaining how preferences fail to capture the thick semantic content of human values, and how utility representations neglect the possible incommensurability of those values. We then critique the normativity of expected utility theory (EUT) for humans and AI, drawing upon arguments showing how rational agents need not comply with EUT, while highlighting how EUT is silent on which preferences are normatively acceptable. Finally, we argue that these limitations motivate a reframing of the targets of AI alignment: Instead of alignment with the preferences of a human user, developer, or humanity-writ-large, AI systems should be aligned with normative standards appropriate to their social roles, such as the role of a general-purpose assistant. Furthermore, these standards should be negotiated and agreed upon by all relevant stakeholders. On this alternative conception of alignment, a multiplicity of AI systems will be able to serve diverse ends, aligned with normative standards that promote mutual benefit and limit harm despite our plural and divergent values.

Improving the alignment of Large Language Models (LLMs) with respect to the cultural values that they encode has become an increasingly important topic. In this work, we study whether we can exploit existing knowledge about cultural values at inference time to adjust model responses to cultural value probes. We present a simple and inexpensive method that uses a combination of in-context learning (ICL) and human survey data, and show that we can improve the alignment to cultural values across 5 models that include both English-centric and multilingual LLMs. Importantly, we show that our method could prove useful in test languages other than English and can improve alignment to the cultural values that correspond to a range of culturally diverse countries.

Numerical weather prediction (NWP) models often underperform compared to simpler climatology-based precipitation forecasts in northern tropical Africa, even after statistical postprocessing. AI-based forecasting models show promise but have avoided precipitation due to its complexity. Synoptic-scale forcings like African easterly waves and other tropical waves (TWs) are important for predictability in tropical Africa, yet their value for predicting daily rainfall remains unexplored. This study uses two machine-learning models--gamma regression and a convolutional neural network (CNN)--trained on TW predictors from satellite-based GPM IMERG data to predict daily rainfall during the July-September monsoon season. Predictor variables are derived from the local amplitude and phase information of seven TW from the target and up-and-downstream neighboring grids at 1-degree spatial resolution. The ML models are combined with Easy Uncertainty Quantification (EasyUQ) to generate calibrated probabilistic forecasts and are compared with three benchmarks: Extended Probabilistic Climatology (EPC15), ECMWF operational ensemble forecast (ENS), and a probabilistic forecast from the ENS control member using EasyUQ (CTRL EasyUQ). The study finds that downstream predictor variables offer the highest predictability, with downstream tropical depression (TD)-type wave-based predictors being most important. Other waves like mixed-Rossby gravity (MRG), Kelvin, and inertio-gravity waves also contribute significantly but show regional preferences. ENS forecasts exhibit poor skill due to miscalibration. CTRL EasyUQ shows improvement over ENS and marginal enhancement over EPC15. Both gamma regression and CNN forecasts significantly outperform benchmarks in tropical Africa. This study highlights the potential of ML models trained on TW-based predictors to improve daily precipitation forecasts in tropical Africa.

Identifying the physical properties of the surrounding environment is essential for robotic locomotion and navigation to deal with non-geometric hazards, such as slippery and deformable terrains. It would be of great benefit for robots to anticipate these extreme physical properties before contact; however, estimating environmental physical parameters from vision is still an open challenge. Animals can achieve this by using their prior experience and knowledge of what they have seen and how it felt. In this work, we propose a cross-modal self-supervised learning framework for vision-based environmental physical parameter estimation, which paves the way for future physical-property-aware locomotion and navigation. We bridge the gap between existing policies trained in simulation and identification of physical terrain parameters from vision. We propose to train a physical decoder in simulation to predict friction and stiffness from multi-modal input. The trained network allows the labeling of real-world images with physical parameters in a self-supervised manner to further train a visual network during deployment, which can densely predict the friction and stiffness from image data. We validate our physical decoder in simulation and the real world using a quadruped ANYmal robot, outperforming an existing baseline method. We show that our visual network can predict the physical properties in indoor and outdoor experiments while allowing fast adaptation to new environments.

The Open Radio Access Network (O-RAN) initiative, characterized by open interfaces and AI/ML-capable RAN Intelligent Controller (RIC), facilitates effective spectrum sharing among RANs. In this context, we introduce AdapShare, an ORAN-compatible solution leveraging Reinforcement Learning (RL) for intent-based spectrum management, with the primary goal of minimizing resource surpluses or deficits in RANs. By employing RL agents, AdapShare intelligently learns network demand patterns and uses them to allocate resources. We demonstrate the efficacy of AdapShare in the spectrum sharing scenario between LTE and NR networks, incorporating real-world LTE resource usage data and synthetic NR usage data to demonstrate its practical use. We use the average surplus or deficit and fairness index to measure the system's performance in various scenarios. AdapShare outperforms a quasi-static resource allocation scheme based on long-term network demand statistics, particularly when available resources are scarce or exceed the aggregate demand from the networks. Lastly, we present a high-level O-RAN compatible architecture using RL agents, which demonstrates the seamless integration of AdapShare into real-world deployment scenarios.

The United States has experienced a significant increase in violent extremism, prompting the need for automated tools to detect and limit the spread of extremist ideology online. This study evaluates the performance of Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-Trained Transformers (GPT) in detecting and classifying online domestic extremist posts. We collected social media posts containing "far-right" and "far-left" ideological keywords and manually labeled them as extremist or non-extremist. Extremist posts were further classified into one or more of five contributing elements of extremism based on a working definitional framework. The BERT model's performance was evaluated based on training data size and knowledge transfer between categories. We also compared the performance of GPT 3.5 and GPT 4 models using different prompts: na\"ive, layperson-definition, role-playing, and professional-definition. Results showed that the best performing GPT models outperformed the best performing BERT models, with more detailed prompts generally yielding better results. However, overly complex prompts may impair performance. Different versions of GPT have unique sensitives to what they consider extremist. GPT 3.5 performed better at classifying far-left extremist posts, while GPT 4 performed better at classifying far-right extremist posts. Large language models, represented by GPT models, hold significant potential for online extremism classification tasks, surpassing traditional BERT models in a zero-shot setting. Future research should explore human-computer interactions in optimizing GPT models for extremist detection and classification tasks to develop more efficient (e.g., quicker, less effort) and effective (e.g., fewer errors or mistakes) methods for identifying extremist content.

The problem of reliably detecting and geolocating objects of different classes in soft real-time is essential in many application areas, such as Search and Rescue performed using Unmanned Aerial Vehicles (UAVs). This research addresses the complementary problems of system contextual vision-based detector selection, allocation, and execution, in addition to the fusion of detection results from teams of UAVs for the purpose of accurately and reliably geolocating objects of interest in a timely manner. In an offline step, an application-independent evaluation of vision-based detectors from a system perspective is first performed. Based on this evaluation, the most appropriate algorithms for online object detection for each platform are selected automatically before a mission, taking into account a number of practical system considerations, such as the available communication links, video compression used, and the available computational resources. The detection results are fused using a method for building maps of salient locations which takes advantage of a novel sensor model for vision-based detections for both positive and negative observations. A number of simulated and real flight experiments are also presented, validating the proposed method.

As language models (LMs) are widely utilized in personalized communication scenarios (e.g., sending emails, writing social media posts) and endowed with a certain level of agency, ensuring they act in accordance with the contextual privacy norms becomes increasingly critical. However, quantifying the privacy norm awareness of LMs and the emerging privacy risk in LM-mediated communication is challenging due to (1) the contextual and long-tailed nature of privacy-sensitive cases, and (2) the lack of evaluation approaches that capture realistic application scenarios. To address these challenges, we propose PrivacyLens, a novel framework designed to extend privacy-sensitive seeds into expressive vignettes and further into agent trajectories, enabling multi-level evaluation of privacy leakage in LM agents' actions. We instantiate PrivacyLens with a collection of privacy norms grounded in privacy literature and crowdsourced seeds. Using this dataset, we reveal a discrepancy between LM performance in answering probing questions and their actual behavior when executing user instructions in an agent setup. State-of-the-art LMs, like GPT-4 and Llama-3-70B, leak sensitive information in 25.68% and 38.69% of cases, even when prompted with privacy-enhancing instructions. We also demonstrate the dynamic nature of PrivacyLens by extending each seed into multiple trajectories to red-team LM privacy leakage risk. Dataset and code are available at https://github.com/SALT-NLP/PrivacyLens.