The spread of fake news has emerged as a critical challenge, undermining trust and posing threats to society. In the era of Large Language Models (LLMs), the capability to generate believable fake content has intensified these concerns. In this study, we present a novel paradigm to evaluate fake news detectors in scenarios involving both human-written and LLM-generated misinformation. Intriguingly, our findings reveal a significant bias in many existing detectors: they are more prone to flagging LLM-generated content as fake news while often misclassifying human-written fake news as genuine. This unexpected bias appears to arise from distinct linguistic patterns inherent to LLM outputs. To address this, we introduce a mitigation strategy that leverages adversarial training with LLM-paraphrased genuine news. The resulting model yielded marked improvements in detection accuracy for both human and LLM-generated news. To further catalyze research in this domain, we release two comprehensive datasets, \texttt{GossipCop++} and \texttt{PolitiFact++}, thus amalgamating human-validated articles with LLM-generated fake and real news.

People use web search engines to find information before forming opinions, which can lead to practical decisions with different levels of impact. The cognitive effort of search can leave opinionated users vulnerable to cognitive biases, e.g., the confirmation bias. In this paper, we investigate whether stance labels and their explanations can help users consume more diverse search results. We automatically classify and label search results on three topics (i.e., intellectual property rights, school uniforms, and atheism) as against, neutral, and in favor, and generate explanations for these labels. In a user study (N =203), we then investigate whether search result stance bias (balanced vs biased) and the level of explanation (plain text, label only, label and explanation) influence the diversity of search results clicked. We find that stance labels and explanations lead to a more diverse search result consumption. However, we do not find evidence for systematic opinion change among users in this context. We believe these results can help designers of search engines to make more informed design decisions.

We introduce a distinctive real-time, causal, neural network-based active speaker detection system optimized for low-power edge computing. This system drives a virtual cinematography module and is deployed on a commercial device. The system uses data originating from a microphone array and a 360-degree camera. Our network requires only 127 MFLOPs per participant, for a meeting with 14 participants. Unlike previous work, we examine the error rate of our network when the computational budget is exhausted, and find that it exhibits graceful degradation, allowing the system to operate reasonably well even in this case. Departing from conventional DOA estimation approaches, our network learns to query the available acoustic data, considering the detected head locations. We train and evaluate our algorithm on a realistic meetings dataset featuring up to 14 participants in the same meeting, overlapped speech, and other challenging scenarios.

This thesis focuses on improving the pre-training of natural language models using unsupervised raw data to make them more efficient and aligned with downstream applications. In the first part, we introduce three alternative pre-training objectives to BERT's Masked Language Modeling (MLM), namely Random Token Substitution (RTS), Cluster-based Random Token Substitution (C-RTS), and Swapped Language Modeling (SLM). These objectives involve token swapping instead of masking, with RTS and C-RTS aiming to predict token originality and SLM predicting the original token values. Results show that RTS and C-RTS require less pre-training time while maintaining performance comparable to MLM. Surprisingly, SLM outperforms MLM on certain tasks despite using the same computational budget. In the second part, we proposes self-supervised pre-training tasks that align structurally with downstream applications, reducing the need for labeled data. We use large corpora like Wikipedia and CC-News to train models to recognize if text spans originate from the same paragraph or document in several ways. By doing continuous pre-training, starting from existing models like RoBERTa, ELECTRA, DeBERTa, BART, and T5, we demonstrate significant performance improvements in tasks like Fact Verification, Answer Sentence Selection, and Summarization. These improvements are especially pronounced when limited annotation data is available. The proposed objectives also achieve state-of-the-art results on various benchmark datasets, including FEVER (dev set), ASNQ, WikiQA, and TREC-QA, as well as enhancing the quality of summaries. Importantly, these techniques can be easily integrated with other methods without altering the internal structure of Transformer models, making them versatile for various NLP applications.

Reward models (RMs) are essential for aligning large language models (LLMs) with human preferences to improve interaction quality. However, the real world is pluralistic, which leads to diversified human preferences with respect to different religions, politics, cultures, etc. Moreover, each individual can have their unique preferences on various topics. Neglecting the diversity of human preferences, current human feedback aligning methods only consider a general reward model, which is below satisfaction for customized or personalized application scenarios. To explore customized preference learning, we collect a domain-specific preference (DSP) dataset, which includes preferred responses for each given query from four practical domains. Besides, from the perspective of data efficiency, we propose a three-stage customized RM learning scheme, then empirically verify its effectiveness on both general preference datasets and our DSP set. Furthermore, we test multiple training and data strategies on the three learning stages. We find several ways to better preserve the general preferring ability while training the customized RMs, especially general preference enrichment, and customized preference imitation learning. The DSP dataset and code are available at https://github.com/Linear95/DSP.

The growing focus on indoor robot navigation utilizing wireless signals has stemmed from the capability of these signals to capture high-resolution angular and temporal measurements. Prior heuristic-based methods, based on radio frequency propagation, are intuitive and generalizable across simple scenarios, yet fail to navigate in complex environments. On the other hand, end-to-end (e2e) deep reinforcement learning (RL), powered by advanced computing machinery, can explore the entire state space, delivering surprising performance when facing complex wireless environments. However, the price to pay is the astronomical amount of training samples, and the resulting policy, without fine-tuning (zero-shot), is unable to navigate efficiently in new scenarios unseen in the training phase. To equip the navigation agent with sample-efficient learning and {zero-shot} generalization, this work proposes a novel physics-informed RL (PIRL) where a distance-to-target-based cost (standard in e2e) is augmented with physics-informed reward shaping. The key intuition is that wireless environments vary, but physics laws persist. After learning to utilize the physics information, the agent can transfer this knowledge across different tasks and navigate in an unknown environment without fine-tuning. The proposed PIRL is evaluated using a wireless digital twin (WDT) built upon simulations of a large class of indoor environments from the AI Habitat dataset augmented with electromagnetic (EM) radiation simulation for wireless signals. It is shown that the PIRL significantly outperforms both e2e RL and heuristic-based solutions in terms of generalization and performance. Source code is available at \url{https://github.com/Panshark/PIRL-WIN}.

Cultural evolution is driven by how we choose what to consume and share with others. A common belief is that the cultural artifacts that succeed are ones that balance novelty and conventionality. This balance theory suggests that people prefer works that are familiar, but not so familiar as to be boring; novel, but not so novel as to violate the expectations of their genre. We test this idea using a large dataset of fanfiction. We apply a multiple regression model and a generalized additive model to examine how the recognition a work receives varies with its novelty, estimated through a Latent Dirichlet Allocation topic model, in the context of existing works. We find the opposite pattern of what the balance theory predicts$\unicode{x2014}$overall success decline almost monotonically with novelty and exhibits a U-shaped, instead of an inverse U-shaped, curve. This puzzle is resolved by teasing out two competing forces: sameness attracts the mass whereas novelty provides enjoyment. Taken together, even though the balance theory holds in terms of expressed enjoyment, the overall success can show the opposite pattern due to the dominant role of sameness to attract the audience. Under these two forces, cultural evolution may have to work against inertia$\unicode{x2014}$the appetite for consuming the familiar$\unicode{x2014}$and may resemble a punctuated equilibrium, marked by occasional leaps.

Artificial intelligence-powered influencers are the new social media trend. But there could be negative effects from the perfect influencers, a humane technologist warns. If you or someone you know is having thoughts of suicide, please contact the Suicide & Crisis Lifeline at 988 or 1-800-273-TALK (8255). Jury selection begins Thursday for the alleged medical abuse case that broke a Florida family apart and inspired the Netflix documentary "Take Care of Maya." In 2016, at just 10 years old, Maya Kowalski -- the girl at the center of the Netflix documentary -- was admitted to Johns Hopkins All Children's Hospital (JHAC) in St. Petersburg, Florida, for severe pain and then promptly removed from the custody of her parents after staff accused them of "medical abuse."

In the film "2001: A Space Odyssey," an astronaut travels through a seeming tunnel of light. Earlier this summer, Artechouse, an organization producing immersive, technology-based art, started offering a science-backed version of a similar trip at its New York venue. The show, titled "Beyond the Light," is a looping twenty-six-minute journey through space and other realms inspired by images from the James Webb Space Telescope (J.W.S.T.). Artechouse began talks with NASA about a show in 2018, and started pulling this one together earlier this year, after the first images captured by J.W.S.T. were released to the public last July. More than sixteen thousand years ago, cave explorers in what's now Lascaux, France, painted animals that are believed to represent the constellations.

Edward Tian didn't think of himself as a writer. As a computer science major at Princeton, he'd taken a couple of journalism classes, where he learned the basics of reporting, and his sunny affect and tinkerer's curiosity endeared him to his teachers and classmates. But he describes his writing style at the time as "pretty bad"--formulaic and clunky. One of his journalism professors said that Tian was good at "pattern recognition," which was helpful when producing news copy. So Tian was surprised when, sophomore year, he managed to secure a spot in John McPhee's exclusive non-fiction writing seminar.