Ukrainian attack drones equipped with artificial intelligence are now finding and attacking targets without human assistance, New Scientist has learned, in what would be the first confirmed use of autonomous weapons or "killer robots". While the drones are designed to target vehicles such as tanks, rather than infantry, it is almost certain that the resulting explosions are killing Russian soldiers without a direct command from a human operator, although no casualties have been confirmed.

The House Armed Services Committee holds a hearing on the Department of Defense using artificial intelligence. Retired Army Gen. Mark Milley believes that artificial intelligence will be a critical component of keeping the U.S. military one step ahead of potential adversaries. "Our military is going to have to change if we are going to continue to be superior to every other military on Earth," Milley, the former chairman of the Joint Chiefs of Staff, said during an interview with "60 Minutes" this week. According to Milley, future wars will look drastically different with the seemingly rapid development of AI technology, something the U.S. will have to be prepared for and adopt if they want to win future wars. The XQ-58A Valkyrie demonstrates the separation of the ALTIUS-600 unmanned aircraft system during a test at the U.S. Army Yuma Proving Ground in Arizona on March 26, 2021.

More than 60 guests--including activists, researchers, policy shapers, and technologists--gathered at the St. Regis San Francisco on Thursday night for a TIME100 Impact Dinner honoring the extraordinary women shaping the future of artificial intelligence. A number of the guests had recently been recognized as leaders in the field by their inclusion in the inaugural TIME100 AI list, which TIME editor in chief Sam Jacobs described as "a map of the relationships and power centers driving the development of AI." TIME CEO Jess Sibley began the evening by speaking further about the philosophy behind the TIME100 AI list. "We looked at the dangers, the perils, but also the power and the progress. We identified 100 people that weren't just Sam Altman, and Reid Hoffman, and Elon Musk, but designers and regulators and researchers. You're going to hear from several of them this evening." Here are some of the biggest moments of the night.

The administration of U.S. President Joe Biden is considering closing a loophole that gives Chinese companies access to American artificial intelligence (AI) chips through units located overseas, according to four people familiar with the matter. The United States last year shook relations with Beijing when it unveiled new restrictions on shipments of AI chips and chipmaking tools to China, seeking to thwart its military advances. Those rules are set to be tightened in the coming days. A person familiar with the situation said the measure could be included in those new restrictions. In the initial round of curbs, the Biden administration left overseas subsidiaries of Chinese companies with unfettered access to the same semiconductors, meaning they could easily be smuggled into China or accessed remotely by China-based users.

Recent global disruptions, such as the pandemic and geopolitical conflicts, have profoundly exposed vulnerabilities in traditional supply chains, requiring exploration of more resilient alternatives. Autonomous supply chains (ASCs) have emerged as a potential solution, offering increased visibility, flexibility, and resilience in turbulent trade environments. Despite discussions in industry and academia over several years, ASCs lack well-established theoretical foundations. This paper addresses this research gap by presenting a formal definition of ASC along with its defining characteristics and auxiliary concepts. We propose a layered conceptual framework called the MIISI model. An illustrative case study focusing on the meat supply chain demonstrates an initial ASC implementation based on this conceptual model. Additionally, we introduce a seven-level supply chain autonomy reference model, delineating a trajectory towards achieving a full supply chain autonomy. Recognising that this work represents an initial endeavour, we emphasise the need for continued exploration in this emerging domain. We anticipate that this work will stimulate further research, both theoretical and technical, and contribute to the continual evolution of ASCs.

Trade restrictions, the COVID-19 pandemic, and geopolitical conflicts has significantly exposed vulnerabilities within traditional global supply chains. These events underscore the need for organisations to establish more resilient and flexible supply chains. To address these challenges, the concept of the autonomous supply chain (ASC), characterised by predictive and self-decision-making capabilities, has recently emerged as promising solution. However, research on ASCs is relatively limited, with no existing studies on their implementations. This paper aims to address this gap by presenting an implementation of ASC using a multi-agent approach. It proposes a methodology for the analysis and design of such an agent-based ASC system (A2SC). This paper provides a concrete case study, the autonomous meat supply chain, which showcases the practical implementation of the A2SC system using the proposed methodology. Additionally, a system architecture and a toolkit for developing A2SC systems are presented. Despite with limitations, this paper demonstrates a promising approach for implementing an effective ASC system.

Patterns of microcircuitry suggest that the brain has an array of repeated canonical computational units. Yet neural representations are distributed, so the relevant computations may only be related indirectly to single-neuron transformations. It thus remains an open challenge how to define canonical distributed computations. We integrate normative and algorithmic theories of neural computation into a mathematical framework for inferring canonical distributed computations from large-scale neural activity patterns. At the normative level, we hypothesize that the brain creates a structured internal model of its environment, positing latent causes that explain its sensory inputs, and uses those sensory inputs to infer the latent causes. At the algorithmic level, we propose that this inference process is a nonlinear message-passing algorithm on a graph-structured model of the world. Given a time series of neural activity during a perceptual inference task, our framework finds (i) the neural representation of relevant latent variables, (ii) interactions between these variables that define the brain's internal model of the world, and (iii) message-functions specifying the inference algorithm. These targeted computational properties are then statistically distinguishable due to the symmetries inherent in any canonical computation, up to a global transformation. As a demonstration, we simulate recordings for a model brain that implicitly implements an approximate inference algorithm on a probabilistic graphical model. Given its external inputs and noisy neural activity, we recover the latent variables, their neural representation and dynamics, and canonical message-functions. We highlight features of experimental design needed to successfully extract canonical computations from neural data. Overall, this framework provides a new tool for discovering interpretable structure in neural recordings.

We study how to perform unlearning, i.e. forgetting undesirable (mis)behaviors, on large language models (LLMs). We show at least three scenarios of aligning LLMs with human preferences can benefit from unlearning: (1) removing harmful responses, (2) erasing copyright-protected content as requested, and (3) eliminating hallucinations. Unlearning, as an alignment technique, has three advantages. (1) It only requires negative (e.g. harmful) examples, which are much easier and cheaper to collect (e.g. via red teaming or user reporting) than positive (e.g. helpful and often human-written) examples required in RLHF (RL from human feedback). (2) It is computationally efficient. (3) It is especially effective when we know which training samples cause the misbehavior. To the best of our knowledge, our work is among the first to explore LLM unlearning. We are also among the first to formulate the settings, goals, and evaluations in LLM unlearning. We show that if practitioners only have limited resources, and therefore the priority is to stop generating undesirable outputs rather than to try to generate desirable outputs, unlearning is particularly appealing. Despite only having negative samples, our ablation study shows that unlearning can still achieve better alignment performance than RLHF with just 2% of its computational time.

Relation extraction aims to classify the relationships between two entities into pre-defined categories. While previous research has mainly focused on sentence-level relation extraction, recent studies have expanded the scope to document-level relation extraction. Traditional relation extraction methods heavily rely on human-annotated training data, which is time-consuming and labor-intensive. To mitigate the need for manual annotation, recent weakly-supervised approaches have been developed for sentence-level relation extraction while limited work has been done on document-level relation extraction. Weakly-supervised document-level relation extraction faces significant challenges due to an imbalanced number "no relation" instances and the failure of directly probing pretrained large language models for document relation extraction. To address these challenges, we propose PromptRE, a novel weakly-supervised document-level relation extraction method that combines prompting-based techniques with data programming. Furthermore, PromptRE incorporates the label distribution and entity types as prior knowledge to improve the performance. By leveraging the strengths of both prompting and data programming, PromptRE achieves improved performance in relation classification and effectively handles the "no relation" problem. Experimental results on ReDocRED, a benchmark dataset for document-level relation extraction, demonstrate the superiority of PromptRE over baseline approaches.

The identification and classification of political claims is an important step in the analysis of political newspaper reports; however, resources for this task are few and far between. This paper explores different strategies for the cross-lingual projection of political claims analysis. We conduct experiments on a German dataset, DebateNet2.0, covering the policy debate sparked by the 2015 refugee crisis. Our evaluation involves two tasks (claim identification and categorization), three languages (German, English, and French) and two methods (machine translation -- the best method in our experiments -- and multilingual embeddings).