OpenAI was reportedly working on an advanced system before Sam Altman's sacking that was so powerful it caused safety concerns among staff at the company. The artificial intelligence model triggered such alarm with some OpenAI researchers that they wrote to the board of directors before Altman's dismissal warning it could threaten humanity, Reuters reported. The model, called Q* – and pronounced as "Q-Star" – was able to solve basic maths problems it had not seen before, according to the tech news site the Information, which added that the pace of development behind the system had alarmed some safety researchers. The ability to solve maths problems would be viewed as a significant development in AI. The reports followed days of turmoil at San Francisco-based OpenAI, whose board sacked Altman last Friday but then reinstated him on Tuesday night after nearly all the company's 750 staff threatened to resign if he was not brought back.

Russian journalist Boris Maksudov has died after sustaining injuries in a drone attack in southeastern Ukraine's Zaporizhia region. Maksudov, who worked for Russian state television Rossiya 24, was wounded on Wednesday and taken to hospital. Initially, defence officials said he was in stable condition. However, he later died of shrapnel wounds. "Boris Maksudov died a hero's death, like a brave fighter," Dmitry Kiselyov, the CEO of the Russian media group Rossia Segodnia, said, according to state-run news agency RIA Novosti.

This week, Emily Bazelon, John Dickerson, and David Plotz discuss the problems with issue polling and issues with political journalism; the chaos and conflict of Sam Altman and OpenAI; and the failure of the Oslo Accords and perpetual struggle between Israel and Palestine. Send us your Conundrums: submit them at slate.com/conundrum. And join us in-person or online with our special guest – The Late Show's Steven Colbert – for Gabfest Live: The Conundrums Edition! December 7 at The 92nd Street Y, New York City. Here are some notes and references from this week's show: Nate Cohn for The New York Times: The Crisis in Issue Polling, and What We're Doing About It and We Did an Experiment to See How Much Democracy and Abortion Matter to Voters Eli Saslow for The New York Times: A Jan. 6 Defendant Pleads His Case to the Son Who Turned Him In John Dickerson and Jo Ling Kent for CBS News Prime Time: What Sam Altman's ouster from OpenAI could mean for the tech world Emily Bazelon for The New York Times Magazine: Was Peace Ever Possible? Ezra Klein for The New York Times's The Ezra Klein Show podcast: The Best Primer I've Heard on Israeli-Palestinian Peace Efforts John Dickerson for CBS Mornings: Former President Jimmy Carter: "America will learn from its mistakes" Here are this week's chatters: John: Julia Simon for NPR: 'It feels like I'm not crazy.'

For a period during the mid-1990s, it was ruled that no blockbuster movie was really complete until it had also been translated into a rock hard platformer or run-and-gun arcade adventure, seemingly designed to enrage and frustrate children everywhere. Disney's wildly uncompromising Aladdin and Lion King tie-ins were shining examples as were Probe Software's challenging Robocop 3 and Alien 3 titles. But veteran Manchester-based publisher Ocean was also a key purveyor. The company spent the 1980s churning out TV and movie games such as Miami Vice, Top Gun and Highlander, but its Jurassic Park titles were among its most ambitious creations and this new collection from cult retro label Limited Run Games brings its NES, SNES and Game Boy translations of the 1993 film together, while also including sequel Jurassic Park 2: The Chaos Continues and two Mega Drive tie-ins created by Bluesky Software: Jurassic Park and Jurassic Park: Rampage Edition. Let's be honest here: none of the games were considered amazing at the time.

Most existing dialogue corpora and models have been designed to fit into 2 predominant categories : task-oriented dialogues portray functional goals, such as making a restaurant reservation or booking a plane ticket, while chit-chat/open-domain dialogues focus on holding a socially engaging talk with a user. However, humans tend to seamlessly switch between modes and even use chitchat to enhance task-oriented conversations. To bridge this gap, new datasets have recently been created, blending both communication modes into conversation examples. The approaches used tend to rely on adding chit-chat snippets to pre-existing, human-generated task-oriented datasets. Given the tendencies observed in humans, we wonder however if the latter do not \textit{already} hold chit-chat sequences. By using topic modeling and searching for topics which are most similar to a set of keywords related to social talk, we explore the training sets of Schema-Guided Dialogues and MultiWOZ. Our study shows that sequences related to social talk are indeed naturally present, motivating further research on ways chitchat is combined into task-oriented dialogues.

In the era of extensive intersection between art and Artificial Intelligence (AI), such as image generation and fiction co-creation, AI for music remains relatively nascent, particularly in music understanding. This is evident in the limited work on deep music representations, the scarcity of large-scale datasets, and the absence of a universal and community-driven benchmark. To address this issue, we introduce the Music Audio Representation Benchmark for universaL Evaluation, termed MARBLE. It aims to provide a benchmark for various Music Information Retrieval (MIR) tasks by defining a comprehensive taxonomy with four hierarchy levels, including acoustic, performance, score, and high-level description. We then establish a unified protocol based on 14 tasks on 8 public-available datasets, providing a fair and standard assessment of representations of all open-sourced pre-trained models developed on music recordings as baselines. Besides, MARBLE offers an easy-to-use, extendable, and reproducible suite for the community, with a clear statement on copyright issues on datasets. Results suggest recently proposed large-scale pre-trained musical language models perform the best in most tasks, with room for further improvement. The leaderboard and toolkit repository are published at https://marble-bm.shef.ac.uk to promote future music AI research.

Current automatic lyrics transcription (ALT) benchmarks focus exclusively on word content and ignore the finer nuances of written lyrics including formatting and punctuation, which leads to a potential misalignment with the creative products of musicians and songwriters as well as listeners' experiences. For example, line breaks are important in conveying information about rhythm, emotional emphasis, rhyme, and high-level structure. To address this issue, we introduce Jam-ALT, a new lyrics transcription benchmark based on the JamendoLyrics dataset. Our contribution is twofold. Firstly, a complete revision of the transcripts, geared specifically towards ALT evaluation by following a newly created annotation guide that unifies the music industry's guidelines, covering aspects such as punctuation, line breaks, spelling, background vocals, and non-word sounds. Secondly, a suite of evaluation metrics designed, unlike the traditional word error rate, to capture such phenomena. We hope that the proposed benchmark contributes to the ALT task, enabling more precise and reliable assessments of transcription systems and enhancing the user experience in lyrics applications such as subtitle renderings for live captioning or karaoke.

Large language models (LLMs) are capable of answering knowledge-intensive complex questions with chain-of-thought (CoT) reasoning. However, they tend to generate factually incorrect reasoning steps when the required knowledge is not available or up-to-date in models' parameters. Recent works turn to retrieving external knowledge to augment CoT reasoning. Despite being promising, these chain-based methods suffer from: 1) Negative retrieval. Unnecessary or incorrect retrieval may mislead the reasoning; 2) Limited sight. Lacking the ability to look backward or forward, a local error in one step will propagate along the chain. In this paper, we propose a novel approach: Probabilistic Tree-of-thought Reasoning (ProbTree). First, LLMs translate a complex question into a query tree, in which each non-root node denotes a sub-question of its parent node. Then, probabilistic reasoning is conducted over the tree, by solving questions from leaf to root considering the confidence of both question decomposing and answering. During reasoning, for leaf nodes, LLMs choose a more confident answer from Closed-book QA that employs parametric knowledge and Open-book QA that employs retrieved external knowledge, thus eliminating the negative retrieval problem. For non-leaf nodes, with the hierarchical structure, LLMs have broader sights and are able to globally reason with the information from child nodes, thus recovering from local errors. The experiments on three Complex QA datasets under the open-domain setting show that our approach outperforms SOTA methods significantly, demonstrating the effect of probabilistic tree-of-thought reasoning.

Heterogeneous graph neural networks (HGNNs) were proposed for representation learning on structural data with multiple types of nodes and edges. To deal with the performance degradation issue when HGNNs become deep, researchers combine metapaths into HGNNs to associate nodes closely related in semantics but far apart in the graph. However, existing metapath-based models suffer from either information loss or high computation costs. To address these problems, we present a novel Metapath Context Convolution-based Heterogeneous Graph Neural Network (MECCH). MECCH leverages metapath contexts, a new kind of graph structure that facilitates lossless node information aggregation while avoiding any redundancy. Specifically, MECCH applies three novel components after feature preprocessing to extract comprehensive information from the input graph efficiently: (1) metapath context construction, (2) metapath context encoder, and (3) convolutional metapath fusion. Experiments on five real-world heterogeneous graph datasets for node classification and link prediction show that MECCH achieves superior prediction accuracy compared with state-of-the-art baselines with improved computational efficiency.

Sam Altman, chief executive officer of OpenAI, during a fireside chat at University College London (UCL) in London, UK, on Wednesday, May 24, 2023. Altman said part of the reason for his current tour of European cities is to discover a suitable location for a new office. ALTMAN RETURNS: OpenAI brings back former CEO, establishes new board days after ouster. EGG ON FACE: OpenAI board's days numbered as $90B company plunges into chaos. NO OVERSIGHT: Tech CEO's ouster demonstrates need for better regulation.