Search

Several federal laws require schools to provide services and accommodations to students with disabilities and to protect them from discrimination. The application of these laws may change depending on students' circumstances. At the P-12 level, three main federal laws protect the rights of students with disabilities: the Individuals with Disabilities Education Act (IDEA), Section 504 of the Rehabilitation Act of 1973 (Section 504), and the Americans with Disabilities Act (ADA). For students receiving special education under the IDEA or accommodations and services under Section 504 or the ADA, transitioning from the P-12 public education system to an institution of higher education (IHE) may affect how a school assesses their disabilities, their eligibility for accommodations or services, and the services and accommodations available to them. Students may also experience different treatment in the same educational context depending on which laws apply to them. This report examines the impact of these laws on students with disabilities in certain key respects: (1) how the laws define disability; (2) how the laws require schools to determine eligibility for services and protections; (3) how the laws ensure students with disabilities receive the accommodations and services they need; (4) the scope of legal protection guaranteed to students with disabilities; (5) how families enforce their rights; and (6) the available remedies.

The properties of the entire neutron star population can be inferred by modeling their evolution, from birth to the present, through pulsar population synthesis. This involves simulating a mock population, applying observational filters, and comparing the resulting sources to the limited subset of detected pulsars. We specifically focus on the magneto-rotational properties of Galactic isolated neutron stars and provide new insights into the intrinsic radio luminosity law by combining pulsar population synthesis with a simulation-based inference (SBI) technique called truncated sequential neural posterior estimation (TSNPE). We employ TSNPE to train a neural density estimator on simulated pulsar populations to approximate the posterior distribution of the underlying parameters. This technique efficiently explores the parameter space by concentrating on regions that are most likely to match the observed data thus allowing a significant reduction in training dataset size. We demonstrate the efficiency of TSNPE over standard neural posterior estimation (NPE), achieving robust inferences of magneto-rotational parameters consistent with previous studies using only around 4% of the simulations required by NPE approaches. Moreover, for the first time, we incorporate data from the Thousand Pulsar Array (TPA) program on MeerKAT, the largest unified sample of neutron stars with consistent fluxes measurement to date, to help constrain the stars' intrinsic radio luminosity. We find that adding flux information as an input to the neural network largely improves the constraints on the pulsars' radio luminosity, as well as improving the estimates on other input parameters.

AIs can beat humans in game environments; however, how helpful those agents are to human remains understudied. We augment CICERO, a natural language agent that demonstrates superhuman performance in Diplomacy, to generate both move and message advice based on player intentions. A dozen Diplomacy games with novice and experienced players, with varying advice settings, show that some of the generated advice is beneficial. It helps novices compete with experienced players and in some instances even surpass them. The mere presence of advice can be advantageous, even if players do not follow it., Comment: 9 pages, 3 figures

Question answering (QA)-producing correct answers for input questions-is popular, but we test a reverse question answering (RQA) task: given an input answer, generate a question with that answer. Past work tests QA and RQA separately, but we test them jointly, comparing their difficulty, aiding benchmark design, and assessing reasoning consistency. 16 LLMs run QA and RQA with trivia questions/answers, showing: 1) Versus QA, LLMs are much less accurate in RQA for numerical answers, but slightly more accurate in RQA for textual answers; 2) LLMs often answer their own invalid questions from RQA accurately in QA, so RQA errors are not from knowledge gaps alone; 3) RQA errors correlate with question difficulty and inversely correlate with answer frequencies in the Dolma corpus; and 4) LLMs struggle to give valid multi-hop questions. By finding question and answer types yielding RQA errors, we suggest improvements for LLM RQA reasoning., Comment: In-progress preprint

We propose a new approach to proving the uniqueness of solutions to a certain class of mean field games of controls. In this class, the equilibrium is determined by an aggregate quantity $Q(t)$, e.g. the market price or production, which then determines optimal trajectories for agents. Our approach consists in analyzing the relationship between $Q(t)$ and corresponding optimal trajectories to find conditions under which there is at most one equilibrium. We show that our conditions do not match those prescribed by the Lasry-Lions monotonicity condition, nor even displacement monotonicity, but they do apply to economic models that have been proposed in the literature.

Recent advancements of large language models (LLMs) have led to claims of AI surpassing humans in natural language processing (NLP) tasks such as textual understanding and reasoning. This work investigates these assertions by introducing CAIMIRA, a novel framework rooted in item response theory (IRT) that enables quantitative assessment and comparison of problem-solving abilities of question-answering (QA) agents: humans and AI systems. Through analysis of over 300,000 responses from ~70 AI systems and 155 humans across thousands of quiz questions, CAIMIRA uncovers distinct proficiency patterns in knowledge domains and reasoning skills. Humans outperform AI systems in knowledge-grounded abductive and conceptual reasoning, while state-of-the-art LLMs like GPT-4 and LLaMA show superior performance on targeted information retrieval and fact-based reasoning, particularly when information gaps are well-defined and addressable through pattern matching or data retrieval. These findings highlight the need for future QA tasks to focus on questions that challenge not only higher-order reasoning and scientific thinking, but also demand nuanced linguistic interpretation and cross-contextual knowledge application, helping advance AI developments that better emulate or complement human cognitive abilities in real-world problem-solving., Comment: To appear at EMNLP 2024 (Main)

Automating the creation of scientific diagrams from academic papers can significantly streamline the development of tutorials, presentations, and posters, thereby saving time and accelerating the process. Current text-to-image models struggle with generating accurate and visually appealing diagrams from long-context inputs. We propose SciDoc2Diagram, a task that extracts relevant information from scientific papers and generates diagrams, along with a benchmarking dataset, SciDoc2DiagramBench. We develop a multi-step pipeline SciDoc2Diagrammer that generates diagrams based on user intentions using intermediate code generation. We observed that initial diagram drafts were often incomplete or unfaithful to the source, leading us to develop SciDoc2Diagrammer-Multi-Aspect-Feedback (MAF), a refinement strategy that significantly enhances factual correctness and visual appeal and outperforms existing models on both automatic and human judgement., Comment: Code and data available at https://github.com/Ishani-Mondal/SciDoc2DiagramGeneration

The dynamics of a neutron star after a glitch involve the transfer of angular momentum from the crust (where the glitch is presumed to originate) to the liquid core, causing the core to spin up. The crust-core coupling, which determines how quickly this spin-up proceeds, can be achieved through various physical processes, including Ekman pumping, superfluid vortex-mediated mutual friction, and magnetic fields. Although the complex nature of these mechanisms has made it difficult to study their combined effects, analytical estimations for individual processes reveal that spin-up timescales vary according to the relative strength of Coriolis, viscous, and mutual friction forces, as well as the magnetic field. However, experimental and numerical validations of those analytical predictions are limited. In this paper, we focus on viscous effects and mutual friction. We conduct non-linear hydrodynamical simulations of the spin-up problem in a two-component fluid by solving the incompressible Hall-Vinen-Bekarevich-Khalatnikov (HVBK) equations in the full sphere (i.e., including $r=0$) for the first time. We find that the viscous (normal) component accelerates due to Ekman pumping, although the mutual friction coupling to the superfluid component alters the spin-up dynamics compared to the single-fluid scenario. Close to the sphere's surface, the response of the superfluid is accurately described by the mutual friction timescale irrespective of its coupling strength with the normal component. However, as we move deeper into the sphere, the superfluid accelerates on different timescales due to the slow viscous spin-up of the internal normal fluid layers. We discuss potential implications for neutron stars and requirements for future work to build more realistic models., Comment: Accepted for Publication in ApJ

Adversarial datasets should ensure AI robustness that matches human performance. However, as models evolve, datasets can become obsolete. Thus, adversarial datasets should be periodically updated based on their degradation in adversarialness. Given the lack of a standardized metric for measuring adversarialness, we propose AdvScore, a human-grounded evaluation metric. AdvScore assesses a dataset's true adversarialness by capturing models' and humans' varying abilities, while also identifying poor examples. AdvScore then motivates a new dataset creation pipeline for realistic and high-quality adversarial samples, enabling us to collect an adversarial question answering (QA) dataset, AdvQA. We apply AdvScore using 9,347 human responses and ten language model predictions to track the models' improvement over five years (from 2020 to 2024). AdvScore assesses whether adversarial datasets remain suitable for model evaluation, measures model improvements, and provides guidance for better alignment with human capabilities., Comment: arXiv admin note: text overlap with arXiv:2401.11185

Keyword mnemonics are memorable explanations that link new terms to simpler keywords. Prior work generates mnemonics for students, but they do not train models using mnemonics students prefer and aid learning. We build SMART, a mnemonic generator trained on feedback from real students learning new terms. To train SMART, we first fine-tune LLaMA-2 on a curated set of user-written mnemonics. We then use LLM alignment to enhance SMART: we deploy mnemonics generated by SMART in a flashcard app to find preferences on mnemonics students favor. We gather 2684 preferences from 45 students across two types: expressed (inferred from ratings) and observed (inferred from student learning), yielding three key findings. First, expressed and observed preferences disagree; what students think is helpful does not always capture what is truly helpful. Second, Bayesian models can synthesize complementary data from multiple preference types into a single effectiveness signal. SMART is tuned via Direct Preference Optimization on this signal, which resolves ties and missing labels in the typical method of pairwise comparisons, augmenting data for LLM output quality gains. Third, mnemonic experts assess SMART as matching GPT-4 at much lower deployment costs, showing the utility of capturing diverse student feedback to align LLMs in education., Comment: EMNLP 2024

Large vision-language models (LVLMs) are prone to hallucinations, where certain contextual cues in an image can trigger the language module to produce overconfident and incorrect reasoning about abnormal or hypothetical objects. While some benchmarks have been developed to investigate LVLM hallucinations, they often rely on hand-crafted corner cases whose failure patterns may not generalize well. Additionally, fine-tuning on these examples could undermine their validity. To address this, we aim to scale up the number of cases through an automated approach, reducing human bias in crafting such corner cases. This motivates the development of AutoHallusion, the first automated benchmark generation approach that employs several key strategies to create a diverse range of hallucination examples. Our generated visual-question pairs pose significant challenges to LVLMs, requiring them to overcome contextual biases and distractions to arrive at correct answers. AutoHallusion enables us to create new benchmarks at the minimum cost and thus overcomes the fragility of hand-crafted benchmarks. It also reveals common failure patterns and reasons, providing key insights to detect, avoid, or control hallucinations. Comprehensive evaluations of top-tier LVLMs, e.g., GPT-4V(ision), Gemini Pro Vision, Claude 3, and LLaVA-1.5, show a 97.7% and 98.7% success rate of hallucination induction on synthetic and real-world datasets of AutoHallusion, paving the way for a long battle against hallucinations. The codebase and data can be accessed at https://github.com/wuxiyang1996/AutoHallusion.

The boardgame Diplomacy is a challenging setting for communicative and cooperative artificial intelligence. The most prominent communicative Diplomacy AI, Cicero, has excellent strategic abilities, exceeding human players. However, the best Diplomacy players master communication, not just tactics, which is why the game has received attention as an AI challenge. This work seeks to understand the degree to which Cicero succeeds at communication. First, we annotate in-game communication with abstract meaning representation to separate in-game tactics from general language. Second, we run two dozen games with humans and Cicero, totaling over 200 human-player hours of competition. While AI can consistently outplay human players, AI-Human communication is still limited because of AI's difficulty with deception and persuasion. This shows that Cicero relies on strategy and has not yet reached the full promise of communicative and cooperative AI.

In this expository article, we give an overview of the concept of potential mean field games of first order. We give a new proof that minimizers of the potential are equilibria by using a Lagrangian formulation. We also provide criteria to determine whether or not a game has a potential. Finally, we discuss in some depth the selection problem in mean field games, which consists in choosing one out of multiple Nash equilibria.

In this article, we study a simplified version of a density-dependent first-order mean field game, in which the players face a penalization equal to the population density at their final position. We consider the problem of finding an equilibrium when the initial distribution is a discrete measure. We show that the problem becomes finite-dimensional: the final piecewise smooth density is completely determined by the weights and positions of the initial measure. We establish existence and uniqueness of a solution using classical fixed point theorems. Finally, we show that Newton's method provides an effective way to compute the solution. Our numerical simulations provide an illustration of how density penalization in a mean field game tends to the smoothen the initial distribution.

Tracking objects in three-dimensional space is critical for autonomous driving. To ensure safety while driving, the tracker must be able to reliably track objects across frames and accurately estimate their states such as velocity and acceleration in the present. Existing works frequently focus on the association task while either neglecting the model performance on state estimation or deploying complex heuristics to predict the states. In this paper, we propose STT, a Stateful Tracking model built with Transformers, that can consistently track objects in the scenes while also predicting their states accurately. STT consumes rich appearance, geometry, and motion signals through long term history of detections and is jointly optimized for both data association and state estimation tasks. Since the standard tracking metrics like MOTA and MOTP do not capture the combined performance of the two tasks in the wider spectrum of object states, we extend them with new metrics called S-MOTA and MOTPS that address this limitation. STT achieves competitive real-time performance on the Waymo Open Dataset., Comment: ICRA 2024

The evolution of artificial intelligence (AI) has catalyzed a transformation in digital content generation, with profound implications for cyber influence operations. This report delves into the potential and limitations of generative deep learning models, such as diffusion models, in fabricating convincing synthetic images. We critically assess the accessibility, practicality, and output quality of these tools and their implications in threat scenarios of deception, influence, and subversion. Notably, the report generates content for several hypothetical cyber influence operations to demonstrate the current capabilities and limitations of these AI-driven methods for threat actors. While generative models excel at producing illustrations and non-realistic imagery, creating convincing photo-realistic content remains a significant challenge, limited by computational resources and the necessity for human-guided refinement. Our exploration underscores the delicate balance between technological advancement and its potential for misuse, prompting recommendations for ongoing research, defense mechanisms, multi-disciplinary collaboration, and policy development. These recommendations aim to leverage AI's potential for positive impact while safeguarding against its risks to the integrity of information, especially in the context of cyber influence., Comment: 44 pages, 56 figures

Flashcard schedulers rely on 1) student models to predict the flashcards a student knows; and 2) teaching policies to pick which cards to show next via these predictions. Prior student models, however, just use study data like the student's past responses, ignoring the text on cards. We propose content-aware scheduling, the first schedulers exploiting flashcard content. To give the first evidence that such schedulers enhance student learning, we build KARL, a simple but effective content-aware student model employing deep knowledge tracing (DKT), retrieval, and BERT to predict student recall. We train KARL by collecting a new dataset of 123,143 study logs on diverse trivia questions. KARL bests existing student models in AUC and calibration error. To ensure our improved predictions lead to better student learning, we create a novel delta-based teaching policy to deploy KARL online. Based on 32 study paths from 27 users, KARL improves learning efficiency over SOTA, showing KARL's strength and encouraging researchers to look beyond historical study data to fully capture student abilities., Comment: EMNLP 2024 (fixed error in testing throughput definition)

Question answering (QA) can only make progress if we know if an answer is correct, but current answer correctness (AC) metrics struggle with verbose, free-form answers from large language models (LLMs). There are two challenges with current short-form QA evaluations: a lack of diverse styles of evaluation data and an over-reliance on expensive and slow LLMs. LLM-based scorers correlate better with humans, but this expensive task has only been tested on limited QA datasets. We rectify these issues by providing rubrics and datasets for evaluating machine QA adopted from the Trivia community. We also propose an efficient, and interpretable QA evaluation that is more stable than an exact match and neural methods(BERTScore)., Comment: Efficient PEDANTS Classifier for short-form QA in github: https://github.com/zli12321/qa_metrics. arXiv admin note: text overlap with arXiv:2401.13170

Bluesky is a new social network built upon the AT Protocol, a decentralized foundation for public social media. It was launched in private beta in February 2023, and has grown to over 10 million registered users by October 2024. In this paper we introduce the architecture of Bluesky and the AT Protocol, and explain how the technical design of Bluesky is informed by our goals: to enable decentralization by having multiple interoperable providers for every part of the system; to make it easy for users to switch providers; to give users agency over the content they see; and to provide a simple user experience that does not burden users with complexity arising from the system's decentralized nature. The system's openness allows anybody to contribute to content moderation and community management, and we invite the research community to use Bluesky as a dataset and testing ground for new approaches in social media moderation.

Topic models are a popular tool for understanding text collections, but their evaluation has been a point of contention. Automated evaluation metrics such as coherence are often used, however, their validity has been questioned for neural topic models (NTMs) and can overlook a models benefits in real world applications. To this end, we conduct the first evaluation of neural, supervised and classical topic models in an interactive task based setting. We combine topic models with a classifier and test their ability to help humans conduct content analysis and document annotation. From simulated, real user and expert pilot studies, the Contextual Neural Topic Model does the best on cluster evaluation metrics and human evaluations; however, LDA is competitive with two other NTMs under our simulated experiment and user study results, contrary to what coherence scores suggest. We show that current automated metrics do not provide a complete picture of topic modeling capabilities, but the right choice of NTMs can be better than classical models on practical task., Comment: 19 pages, 5 tables, 6 figures, Accepted to EACL Main Conference 2024

Neutron stars are compact and dense celestial objects that offer the unique opportunity to explore matter and its interactions under conditions that cannot be reproduced elsewhere in the Universe. Their extreme gravitational, rotational and magnetic energy reservoirs fuel the large variety of their emission, which encompasses all available multi-messenger tracers: electromagnetic and gravitational waves, neutrinos, and cosmic rays. However, accurately measuring global neutron-star properties such as mass, radius, and moment of inertia poses significant challenges. Probing internal characteristics such as the crustal composition or superfluid physics is even more complex. This article provides a comprehensive review of the different methods employed to measure neutron-star characteristics and the level of reliance on theoretical models. Understanding these measurement techniques is crucial for advancing our knowledge of neutron-star physics. We also highlight the importance of employing independent methods and adopting a multi-messenger approach to gather complementary data from various observable phenomena as exemplified by the recent breakthroughs in gravitational-wave astronomy and the landmark detection of a binary neutron-star merger. Consolidating the current state of knowledge on neutron-star measurements will enable an accurate interpretation of the current data and errors, and better planning for future observations and experiments., Comment: 48 pages (including 11 pages of references), 10 figures, 2 tables; invited review for special issue of Astroparticle Physics on 'Gravitational Waves and Multi-messenger Astrophysics'

Question answering (QA) can only make progress if we know if an answer is correct, but for many of the most challenging and interesting QA examples, current evaluation metrics to determine answer equivalence (AE) often do not align with human judgments, particularly more verbose, free-form answers from large language models (LLM). There are two challenges: a lack of data and that models are too big: LLM-based scorers can correlate better with human judges, but this task has only been tested on limited QA datasets, and even when available, update of the model is limited because LLMs are large and often expensive. We rectify both of these issues by providing clear and consistent guidelines for evaluating AE in machine QA adopted from professional human QA contests. We also introduce a combination of standard evaluation and a more efficient, robust, and lightweight discriminate AE classifier-based matching method (CFMatch, smaller than 1 MB), trained and validated to more accurately evaluate answer correctness in accordance with adopted expert AE rules that are more aligned with human judgments., Comment: A duplicate and polished version is in arXiv:2402.11161

Dynamic adversarial question generation, where humans write examples to stump a model, aims to create examples that are realistic and informative. However, the advent of large language models (LLMs) has been a double-edged sword for human authors: more people are interested in seeing and pushing the limits of these models, but because the models are so much stronger an opponent, they are harder to defeat. To understand how these models impact adversarial question writing process, we enrich the writing guidance with LLMs and retrieval models for the authors to reason why their questions are not adversarial. While authors could create interesting, challenging adversarial questions, they sometimes resort to tricks that result in poor questions that are ambiguous, subjective, or confusing not just to a computer but also to humans. To address these issues, we propose new metrics and incentives for eliciting good, challenging questions and present a new dataset of adversarially authored questions.

We combine pulsar population synthesis with simulation-based inference (SBI) to constrain the magnetorotational properties of isolated Galactic radio pulsars. We first develop a framework to model neutron star birth properties and their dynamical and magnetorotational evolution. We specifically sample initial magnetic field strengths, $B$, and spin periods, $P$, from lognormal distributions and capture the late-time magnetic field decay with a power law. Each lognormal is described by a mean, $\mu_{\log B}, \mu_{\log P}$, and standard deviation, $\sigma_{\log B}, \sigma_{\log P}$, while the power law is characterized by the index, $a_{\rm late}$. We subsequently model the stars' radio emission and observational biases to mimic detections with three radio surveys, and we produce a large database of synthetic $P$--$\dot{P}$ diagrams by varying our five magnetorotational input parameters. We then follow an SBI approach that focuses on neural posterior estimation and train deep neural networks to infer the parameters' posterior distributions. After successfully validating these individual neural density estimators on simulated data, we use an ensemble of networks to infer the posterior distributions for the observed pulsar population. We obtain $\mu_{\log B} = 13.10^{+0.08}_{-0.10}$, $\sigma_{\log B} = 0.45^{+0.05}_{-0.05}$ and $\mu_{\log P} = -1.00^{+0.26}_{-0.21}$, $\sigma_{\log P} = 0.38^{+0.33}_{-0.18}$ for the lognormal distributions and $a_{\rm late} = -1.80^{+0.65}_{-0.61}$ for the power law at the $95\%$ credible interval. We contrast our results with previous studies and highlight uncertainties of the inferred $a_{\rm late}$ value. Our approach represents a crucial step toward robust statistical inference for complex population synthesis frameworks and forms the basis for future multiwavelength analyses of Galactic pulsars., Comment: 31 pages, 16 figures, 5 tables, 2 appendices; published version

Translations help people understand content written in another language. However, even correct literal translations do not fulfill that goal when people lack the necessary background to understand them. Professional translators incorporate explicitations to explain the missing context by considering cultural differences between source and target audiences. Despite its potential to help users, NLP research on explicitation is limited because of the dearth of adequate evaluation methods. This work introduces techniques for automatically generating explicitations, motivated by WikiExpl: a dataset that we collect from Wikipedia and annotate with human translators. The resulting explicitations are useful as they help answer questions more accurately in a multilingual question answering framework., Comment: EMNLP2023

Questions posed by information-seeking users often contain implicit false or potentially harmful assumptions. In a high-risk domain such as maternal and infant health, a question-answering system must recognize these pragmatic constraints and go beyond simply answering user questions, examining them in context to respond helpfully. To achieve this, we study assumptions and implications, or pragmatic inferences, made when mothers ask questions about pregnancy and infant care by collecting a dataset of 2,727 inferences from 500 questions across three diverse sources. We study how health experts naturally address these inferences when writing answers, and illustrate that informing existing QA pipelines with pragmatic inferences produces responses that are more complete, mitigating the propagation of harmful beliefs., Comment: Accepted to NAACL 2024

Topic models are valuable for understanding extensive document collections, but they don't always identify the most relevant topics. Classical probabilistic and anchor-based topic models offer interactive versions that allow users to guide the models towards more pertinent topics. However, such interactive features have been lacking in neural topic models. To correct this lacuna, we introduce a user-friendly interaction for neural topic models. This interaction permits users to assign a word label to a topic, leading to an update in the topic model where the words in the topic become closely aligned with the given label. Our approach encompasses two distinct kinds of neural topic models. The first includes models where topic embeddings are trainable and evolve during the training process. The second kind involves models where topic embeddings are integrated post-training, offering a different approach to topic refinement. To facilitate user interaction with these neural topic models, we have developed an interactive interface. This interface enables users to engage with and re-label topics as desired. We evaluate our method through a human study, where users can relabel topics to find relevant documents. Using our method, user labeling improves document rank scores, helping to find more relevant documents to a given query when compared to no user labeling.

We study a simple first-order mean field game in which the coupling with the mean field is only in the final time and gives an incentive for players to congregate. For a short enough time horizon, the equilibrium is unique. We consider the process of \emph{iterating} the game, taking the final population distribution as the initial distribution in the next iteration. Restricting to one dimension, we take this to be a model of coalition building for a population distributed over some spectrum of opinions. Our main result states that, given a final coupling of the form $G(x,m) = \int \varphi(x-z)\dif m(z)$ where $\varphi$ is a smooth, even, non-positive function of compact support, then as the number of iterations goes to infinity the population tends to cluster into discrete groups, which are spread out as a function of the size of the support of $\varphi$. We discuss the potential implications of this result for real-world opinion dynamics and political systems., Comment: 22 pages, 4 figures, Submitted to Dynamic Games and Applications

Large Language Models (LLMs) are deployed in interactive contexts with direct user engagement, such as chatbots and writing assistants. These deployments are vulnerable to prompt injection and jailbreaking (collectively, prompt hacking), in which models are manipulated to ignore their original instructions and follow potentially malicious ones. Although widely acknowledged as a significant security threat, there is a dearth of large-scale resources and quantitative studies on prompt hacking. To address this lacuna, we launch a global prompt hacking competition, which allows for free-form human input attacks. We elicit 600K+ adversarial prompts against three state-of-the-art LLMs. We describe the dataset, which empirically verifies that current LLMs can indeed be manipulated via prompt hacking. We also present a comprehensive taxonomical ontology of the types of adversarial prompts., Comment: 34 pages, 8 figures Codebase: https://github.com/PromptLabs/hackaprompt Dataset: https://huggingface.co/datasets/hackaprompt/hackaprompt-dataset/blob/main/README.md Playground: https://huggingface.co/spaces/hackaprompt/playground

Polarization and the marketplace for impressions have conspired to make navigating information online difficult for users, and while there has been a significant effort to detect false or misleading text, multimodal datasets have received considerably less attention. To complement existing resources, we present multimodal Video Misleading Headline (VMH), a dataset that consists of videos and whether annotators believe the headline is representative of the video's contents. After collecting and annotating this dataset, we analyze multimodal baselines for detecting misleading headlines. Our annotation process also focuses on why annotators view a video as misleading, allowing us to better understand the interplay of annotators' background and the content of the videos., Comment: EMNLP 2023 Main Paper

Large Language Models (LLMs) are increasingly used for accessing information on the web. Their truthfulness and factuality are thus of great interest. To help users make the right decisions about the information they get, LLMs should not only provide information but also help users fact-check it. Our experiments with 80 crowdworkers compare language models with search engines (information retrieval systems) at facilitating fact-checking. We prompt LLMs to validate a given claim and provide corresponding explanations. Users reading LLM explanations are significantly more efficient than those using search engines while achieving similar accuracy. However, they over-rely on the LLMs when the explanation is wrong. To reduce over-reliance on LLMs, we ask LLMs to provide contrastive information - explain both why the claim is true and false, and then we present both sides of the explanation to users. This contrastive explanation mitigates users' over-reliance on LLMs, but cannot significantly outperform search engines. Further, showing both search engine results and LLM explanations offers no complementary benefits compared to search engines alone. Taken together, our study highlights that natural language explanations by LLMs may not be a reliable replacement for reading the retrieved passages, especially in high-stakes settings where over-relying on wrong AI explanations could lead to critical consequences., Comment: NAACL 2024