Bayesian active learning is based on information theoretical approaches that focus on maximising the information that new observations provide to the model parameters. This is commonly done by maximising the Bayesian Active Learning by Disagreement (BALD) acquisitions function. However, we highlight that it is challenging to estimate BALD when the new data points are subject to censorship, where only clipped values of the targets are observed. To address this, we derive the entropy and the mutual information for censored distributions and derive the BALD objective for active learning in censored regression ($\mathcal{C}$-BALD). We propose a novel modelling approach to estimate the $\mathcal{C}$-BALD objective and use it for active learning in the censored setting. Across a wide range of datasets and models, we demonstrate that $\mathcal{C}$-BALD outperforms other Bayesian active learning methods in censored regression.

When Ashley Neale started college in Texas in 2013, she needed money to pay for school. So, at the age of 18, she worked first as a cam girl and then as a stripper. Men would try to slip their fingers between her legs as she walked from the stage to the dressing room so often that she learned how to dislocate their shoulders. After her third successful dislocation, her manager told her to stop defending herself. Since then, she's continued her career in sex work – but in the tech world. She worked at FetLife, a social network for the fetish community; experimented with a subscription site for adult content where users paid in crypto; and has now created her own AI romance app: MyPeach.ai, which uses AI-generated text and imagery to replicate the experience of chatting – and sexting – with someone online.

Large Language Models (LLMs), such as ChatGPT and GPT-4, are designed to provide useful and safe responses. However, adversarial prompts known as 'jailbreaks' can circumvent safeguards, leading LLMs to generate potentially harmful content. Exploring jailbreak prompts can help to better reveal the weaknesses of LLMs and further steer us to secure them. Unfortunately, existing jailbreak methods either suffer from intricate manual design or require optimization on other white-box models, compromising generalization or efficiency. In this paper, we generalize jailbreak prompt attacks into two aspects: (1) Prompt Rewriting and (2) Scenario Nesting. Based on this, we propose ReNeLLM, an automatic framework that leverages LLMs themselves to generate effective jailbreak prompts. Extensive experiments demonstrate that ReNeLLM significantly improves the attack success rate while greatly reducing the time cost compared to existing baselines. Our study also reveals the inadequacy of current defense methods in safeguarding LLMs. Finally, we analyze the failure of LLMs defense from the perspective of prompt execution priority, and propose corresponding defense strategies. We hope that our research can catalyze both the academic community and LLMs developers towards the provision of safer and more regulated LLMs. The code is available at https://github.com/NJUNLP/ReNeLLM.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. The Idaho legislature passed a bill this week to carry out the death penalty for sex crimes against children younger than 12. Another bill permitting prosecutors to bring sexual exploitation charges against producers of child pornography using artificial intelligence (AI) also passed the assembly in the same session. HB 515 would amend Idaho's current statute that carries a life sentence for "lewd conduct with a minor" below the age of 16. If the child is under 12, if the act is "especially heinous, atrocious or cruel, manifesting exceptional depravity," then prosecutors would seek the death penalty.

Many of us have had bad experiences of being swiped left, ghosted, breadcrumbed and benched on internet dating apps – though few people have ever thought to take their heartbreak to court. On Valentine's Day, six dating app users filed a proposed class-action lawsuit accusing Tinder, Hinge and other Match dating apps of using addictive, game-like features to encourage compulsive use. Match's apps, according to the lawsuit filed in federal court in the Northern District of California, "employ recognised dopamine-manipulating product features" to turn users into "gamblers locked in a search for psychological rewards", generating "market success by fomenting dating app addiction that drives expensive subscriptions and perpetual use". Match said the lawsuit was "ridiculous", but online dating experts said it reflected a broader backlash to the way apps were gamifying human experience for profit and leaving people feeling manipulated. "I'm not at all surprised that this has come to litigation. I think big tech is the new big tobacco, as smartphones are just as addictive as cigarettes," said Mia Levitin, author of The Future of Seduction.

This survey paper presents a comprehensive analysis of crime prediction methodologies, exploring the various techniques and technologies utilized in this area. The paper covers the statistical methods, machine learning algorithms, and deep learning techniques employed to analyze crime data, while also examining their effectiveness and limitations. We propose a methodological taxonomy that classifies crime prediction algorithms into specific techniques. This taxonomy is structured into four tiers, including methodology category, methodology sub-category, methodology techniques, and methodology sub-techniques. Empirical and experimental evaluations are provided to rank the different techniques. The empirical evaluation assesses the crime prediction techniques based on four criteria, while the experimental evaluation ranks the algorithms that employ the same sub-technique, the different sub-techniques that employ the same technique, the different techniques that employ the same methodology sub-category, the different methodology sub-categories within the same category, and the different methodology categories. The combination of methodological taxonomy, empirical evaluations, and experimental comparisons allows for a nuanced and comprehensive understanding of crime prediction algorithms, aiding researchers in making informed decisions. Finally, the paper provides a glimpse into the future of crime prediction techniques, highlighting potential advancements and opportunities for further research in this field

Dating apps are supposedly'designed to be deleted,' but a new class action lawsuit claims the apps are instead'designed to be addictive.' The lawsuit, filed on Valentine's Day against Match Group which owns Tinder, Hinge, Match, OkCupid, and Plenty of Fish, accused the company of using'psychological manipulation' like push notifications, rewards, and punishments to guarantee users keep swiping right. The app is designed to turn users into'addicts' who are enticed by the game-like play-to-play loop, the lawsuit claimed, accusing Match Group of prioritizing profit over promises to help users find love. Match sells subscription plans to remove like limits and see who likes you with Tinder offering its Gold package for 140 for six months or 40 for one month and its Platinum package for 50 per month or 180 for six months. The lawsuit claims that if users were content with the basic app features, they wouldn't need to purchase the additional subscription when they reach their'like limit.'

The promise of artificial intelligence in medicine is that it can reduce the influence of human error and bias in health care. But there's growing concern that A.I. in medicine –as in other fields– can reflect the biases and lack of diversity among its creators. And that can have life threatening consequences for African American patients. On today's episode of A Word, Jason Johnson is joined by Margo Snipe, a health reporter for CapitalB News. They discuss how A.I. can sometimes fuel medical racism, and reasons to hope that it can change.

On July 20, 2023, a group of 27 scholars and digital rights advocates with expertise in law, computer science, political science, and other disciplines gathered for the Large Language Models, Law and Policy Roundtable, co-hosted by the NYU School of Law's Information Law Institute and the Center for Democracy & Technology. The roundtable convened to discuss how law and policy can help address some of the larger societal problems posed by large language models (LLMs). The discussion focused on three policy topic areas in particular: 1. Truthfulness: What risks do LLMs pose in terms of generating mis- and disinformation? How can these risks be mitigated from a technical and/or regulatory perspective? 2. Privacy: What are the biggest privacy risks involved in the creation, deployment, and use of LLMs? How can these risks be mitigated from a technical and/or regulatory perspective? 3. Market concentration: What threats do LLMs pose concerning market/power concentration? How can these risks be mitigated from a technical and/or regulatory perspective? In this paper, we provide a detailed summary of the day's proceedings. We first recap what we deem to be the most important contributions made during the issue framing discussions. We then provide a list of potential legal and regulatory interventions generated during the brainstorming discussions.

Several parameter-efficient fine-tuning methods based on adapters have been proposed as a streamlined approach to incorporate not only a single specialized knowledge into existing Pre-Trained Language Models (PLMs) but also multiple of them at once. Recent works such as AdapterSoup propose to mix not all but only a selective sub-set of domain-specific adapters during inference via model weight averaging to optimize performance on novel, unseen domains with excellent computational efficiency. However, the essential generalizability of this emerging weight-space adapter mixing mechanism on unseen, in-domain examples remains unexplored. Thus, in this study, we conduct a comprehensive analysis to elucidate the generalizability of domain-specific adapter mixtures in in-domain evaluation. We also provide investigations into the inner workings of the mixture of domain-specific adapters by analyzing their weight signs, yielding critical analysis on the negative correlation between their fraction of weight sign difference and their mixtures' generalizability. All source code will be published.