Called "Tokyo Futari Story," the city hall's new initiative is just that: An effort to create couples, "futari," in a country where it is increasingly common to be "hitori," or alone. While a site offering counsel and general information for potential lovebirds is online, a dating app is also in development. City hall hopes to offer it later this year, accessible through phone or web, a city official said Thursday. City Hall declined to comment on Japanese media reports that said the app will require a confirmation of identity, such as a driver's license, your tax records to prove income and a signed form that says you are ready to get married. 'MEET HOT, SINGLE FIREMEN, SCORE A PRIZE': NEWEST WAY WOMEN ARE FINDING THEIR LOVE MATCHES Marriage is on the decline in Japan as the country's birth rate fell to an all-time low, according to health ministry data on Wednesday.

NewsBreak, a popular free news app in the US, has been publishing fictitious stories written by AI since 2021, according to Reuters. The app publishes licensed content from legitimate news sources, such as CNN, AP and Reuters itself, but it also uses artificial intelligence tools to rewrite press releases and local news. One of the most egregious examples of a false news story by NewsBreak was published on Christmas Eve last year. The app's writeup claimed that there was a shooting in Bridgeton, New Jersey when no such incident took place. New Jersey's police department dismissed the claims made in the article before the app, which said it got the information from another website, took it down four days later.

The cartoon interviewer greets you on screen. He looks a little young to be asking questions about a job--sort of a cartoon version of Harry Potter, with dark hair and glasses. You can choose other interviewers to speak with instead, representing various genders and races with names like Benjamin, Leslie, and Kristin. Alex, the name given to this AI interviewer, asks about your professional experience, theoretical questions about programming, and then gives out a coding exercise. Alex is an AI interviewer developed by micro1, a US company that describes itself as an AI recruitment engine for engineers.

Ten years ago, Eliot Higgins could eat room service meals at a hotel without fear of being poisoned. He hadn't yet been declared a foreign agent by Russia; in fact, he wasn't even a blip on the radar of security agencies in that country or anywhere else. He was just a British guy with an unfulfilling admin job who'd been blogging under the pen name Brown Moses--after a Frank Zappa song--and was in the process of turning his blog into a full-fledged website. He was an open source intelligence analyst avant la lettre, poring over social media photos and videos and other online jetsam to investigate wartime atrocities in Libya and Syria. In its disorganized way, the internet supplied him with so much evidence that he was beating UN investigators to their conclusions.

Before being deployed for user-facing applications, developers align Large Language Models (LLMs) to user preferences through a variety of procedures, such as Reinforcement Learning From Human Feedback (RLHF) and Direct Preference Optimization (DPO). Current evaluations of these procedures focus on benchmarks of instruction following, reasoning, and truthfulness. However, human preferences are not universal, and aligning to specific preference sets may have unintended effects. We explore how alignment impacts performance along three axes of global representation: English dialects, multilingualism, and opinions from and about countries worldwide. Our results show that current alignment procedures create disparities between English dialects and global opinions. We find alignment improves capabilities in several languages. We conclude by discussing design decisions that led to these unintended impacts and recommendations for more equitable preference tuning. We make our code and data publicly available on Github.

Retrieval-Augmented Generation (RAG) is an effective solution to supplement necessary knowledge to large language models (LLMs). Targeting its bottleneck of retriever performance, "generate-then-read" pipeline is proposed to replace the retrieval stage with generation from the LLM itself. Although promising, this research direction is underexplored and still cannot work in the scenario when source knowledge is given. In this paper, we formalize a general "A + B" framework with varying combinations of foundation models and types for systematic investigation. We explore the efficacy of the base and chat versions of LLMs and found their different functionalities suitable for generator A and reader B, respectively. Their combinations consistently outperform single models, especially in complex scenarios. Furthermore, we extend the application of the "A + B" framework to scenarios involving source documents through continuous learning, enabling the direct integration of external knowledge into LLMs. This approach not only facilitates effective acquisition of new knowledge but also addresses the challenges of safety and helpfulness post-adaptation. The paper underscores the versatility of the "A + B" framework, demonstrating its potential to enhance the practical application of LLMs across various domains.

Large Language Models (LLMs) have achieved state-of-the-art performance at zero-shot generation of abstractive summaries for given articles. However, little is known about the robustness of such a process of zero-shot summarization. To bridge this gap, we propose relevance paraphrasing, a simple strategy that can be used to measure the robustness of LLMs as summarizers. The relevance paraphrasing approach identifies the most relevant sentences that contribute to generating an ideal summary, and then paraphrases these inputs to obtain a minimally perturbed dataset. Then, by evaluating model performance for summarization on both the original and perturbed datasets, we can assess the LLM's one aspect of robustness. We conduct extensive experiments with relevance paraphrasing on 4 diverse datasets, as well as 4 LLMs of different sizes (GPT-3.5-Turbo, Llama-2-13B, Mistral-7B, and Dolly-v2-7B). Our results indicate that LLMs are not consistent summarizers for the minimally perturbed articles, necessitating further improvements.

The increasing complexity of foundational models underscores the necessity for explainability, particularly for fine-tuning, the most widely used training method for adapting models to downstream tasks. Instance attribution, one type of explanation, attributes the model prediction to each training example by an instance score. However, the robustness of instance scores, specifically towards dataset resampling, has been overlooked. To bridge this gap, we propose a notion of robustness on the sign of the instance score. We theoretically and empirically demonstrate that the popular leave-one-out-based methods lack robustness, while the Shapley value behaves significantly better, but at a higher computational cost. Accordingly, we introduce an efficient fine-tuning-free approximation of the Shapley value (FreeShap) for instance attribution based on the neural tangent kernel. We empirically demonstrate that FreeShap outperforms other methods for instance attribution and other data-centric applications such as data removal, data selection, and wrong label detection, and further generalize our scale to large language models (LLMs). Our code is available at https://github.com/JTWang2000/FreeShap.

Next-item recommender systems are often trained using only positive feedback with randomly-sampled negative feedback. We show the benefits of using real negative feedback both as inputs into the user sequence and also as negative targets for training a next-song recommender system for internet radio. In particular, using explicit negative samples during training helps reduce training time by ~60% while also improving test accuracy by ~6%; adding user skips as additional inputs also can considerably increase user coverage alongside slightly improving accuracy. We test the impact of using a large number of random negative samples to capture a 'harder' one and find that the test accuracy increases with more randomly-sampled negatives, but only to a point. Too many random negatives leads to false negatives that limits the lift, which is still lower than if using true negative feedback. We also find that the test accuracy is fairly robust with respect to the proportion of different feedback types, and compare the learned embeddings for different feedback types.

The explainability of recommender systems has attracted significant attention in academia and industry. Many efforts have been made for explainable recommendations, yet evaluating the quality of the explanations remains a challenging and unresolved issue. In recent years, leveraging LLMs as evaluators presents a promising avenue in Natural Language Processing tasks (e.g., sentiment classification, information extraction), as they perform strong capabilities in instruction following and common-sense reasoning. However, evaluating recommendation explanatory texts is different from these NLG tasks, as its criteria are related to human perceptions and are usually subjective. In this paper, we investigate whether LLMs can serve as evaluators of recommendation explanations. To answer the question, we utilize real user feedback on explanations given from previous work and additionally collect third-party annotations and LLM evaluations. We design and apply a 3-level meta evaluation strategy to measure the correlation between evaluator labels and the ground truth provided by users. Our experiments reveal that LLMs, such as GPT4, can provide comparable evaluations with appropriate prompts and settings. We also provide further insights into combining human labels with the LLM evaluation process and utilizing ensembles of multiple heterogeneous LLM evaluators to enhance the accuracy and stability of evaluations. Our study verifies that utilizing LLMs as evaluators can be an accurate, reproducible and cost-effective solution for evaluating recommendation explanation texts. Our code is available at https://github.com/Xiaoyu-SZ/LLMasEvaluator.