An optimal delivery of arguments is key to persuasion in any debate, both for humans and for AI systems. This requires the use of clear and fluent claims relevant to the given debate. Prior work has studied the automatic assessment of argument quality extensively. Yet, no approach actually improves the quality so far. To fill this gap, this paper proposes the task of claim optimization: to rewrite argumentative claims in order to optimize their delivery. As multiple types of optimization are possible, we approach this task by first generating a diverse set of candidate claims using a large language model, such as BART, taking into account contextual information. Then, the best candidate is selected using various quality metrics. In automatic and human evaluation on an English-language corpus, our quality-based candidate selection outperforms several baselines, improving 60% of all claims (worsening 16% only). Follow-up analyses reveal that, beyond copy editing, our approach often specifies claims with details, whereas it adds less evidence than humans do. Moreover, its capabilities generalize well to other domains, such as instructional texts.

We present a novel optimization algorithm, element-wise relaxed scalar auxiliary variable (E-RSAV), that satisfies an unconditional energy dissipation law and exhibits improved alignment between the modified and the original energy. Our algorithm features rigorous proofs of linear convergence in the convex setting. Furthermore, we present a simple accelerated algorithm that improves the linear convergence rate to super-linear in the univariate case. We also propose an adaptive version of E-RSAV with Steffensen step size.

Accurate predictions, as with machine learning, may not suffice to provide optimal healthcare for every patient. Indeed, prediction can be driven by shortcuts in the data, such as racial biases. Causal thinking is needed for data-driven decisions. Here, we give an introduction to the key elements, focusing on routinely-collected data, electronic health records (EHRs) and claims data. Using such data to assess the value of an intervention requires care: temporal dependencies and existing practices easily confound the causal effect. We present a step-by-step framework to help build valid decision making from real-life patient records by emulating a randomized trial before individualizing decisions, eg with machine learning. Our framework highlights the most important pitfalls and considerations in analysing EHRs or claims data to draw causal conclusions. We illustrate the various choices in studying the effect of albumin on sepsis mortality in the Medical Information Mart for Intensive Care database (MIMIC-IV). We study the impact of various choices at every step, from feature extraction to causal-estimator selection. In a tutorial spirit, the code and the data are openly available.

As input data distributions evolve, the predictive performance of machine learning models tends to deteriorate. In practice, new input data tend to come without target labels. Then, state-of-the-art techniques model input data distributions or model prediction distributions and try to understand issues regarding the interactions between learned models and shifting distributions. We suggest a novel approach that models how explanation characteristics shift when affected by distribution shifts. We find that the modeling of explanation shifts can be a better indicator for detecting out-of-distribution model behaviour than state-of-the-art techniques. We analyze different types of distribution shifts using synthetic examples and real-world data sets. We provide an algorithmic method that allows us to inspect the interaction between data set features and learned models and compare them to the state-of-the-art. We release our methods in an open-source Python package, as well as the code used to reproduce our experiments.

AI-generated images and audio are already making their way into the 2024 Presidential election cycle. In an effort to staunch the flow of disinformation ahead of what is expected to be a contentious election, Google announced on Wednesday that it will require political advertisers to "prominently disclose" whenever their advertisement contains AI-altered or -generated aspects, "inclusive of AI tools." The new rules will based on the company's existing Manipulated Media Policy and will take effect in November. "Given the growing prevalence of tools that produce synthetic content, we're expanding our policies a step further to require advertisers to disclose when their election ads include material that's been digitally altered or generated," a Google spokesperson said in a statement obtained by The Hill. Small and inconsequential edits like resizing images, minor cleanup to the background or color correction will all still be allowed -- those that depict people or things doing stuff that they never actually did or those that otherwise alter actual footage will be flagged.

The home state of some of the most influential AI companies has a new plan to confront the potential regulation of generative AI. California Governor Gavin Newsom signed an executive order instructing agencies in the state to study potential risks and use cases for the technology. Under the order, state agencies are tasked with identifying "the most significant and beneficial uses of GenAI in the state" and creating frameworks to train state employees on how to use "state-approved" generative AI tools in their work. Likewise, it directs the same agencies to analyze potential negative impacts of the technology, including its effect on vulnerable communities and threats to "critical energy infrastructure" in the state. The order also lays the groundwork for new partnerships with University of California at Berkeley and Stanford University, which will help study how generative AI is affecting the state's workers.

California Gov. Gavin Newsom on Wednesday signed an executive order directing state agencies to examine the benefits and risks of artificial intelligence that can generate text, images and other content. The executive order sets the stage for potential regulation around what's known as generative AI technology, which has already raised concerns about misinformation, plagiarism, bias and child safety. The governor and California lawmakers thus far have been cautious about regulating technology they might not fully understand and hindering business innovations that fuel the state's economy. "We recognize both the potential benefits and risks these tools enable. We're neither frozen by the fears nor hypnotized by the upside," Newsom said in a statement.

United States, British and European Union representatives are visiting the United Arab Emirates amid concerns regarding shipments of goods, including computer chips, to Russia that could help Moscow in its war on Ukraine. The senior Western officials arrived in the Gulf nation this week to discuss sanctions on Russia, as concerns mounted that Moscow was bypassing them through various means, a US embassy spokesperson told CNN on Wednesday. The report came on the heels of another by the Wall Street Journal on Monday – citing US and European officials – that discussed plans to jointly press the UAE to halt shipments of goods to Russia. This was part of a collective global push to keep computer chips, electronic components and other so-called dual-use products out of Russian hands, the WSJ report said. The UAE, a member of the OPEC oil alliance that includes Russia, has maintained good ties with Moscow despite Western pressure to isolate Russia over the invasion of Ukraine that began in February 2022.

Parts of what could be a Russian drone fell on Romanian territory, Romania's Defence Minister Angel Tilvar says, two days after Ukraine said Russian drones had detonated on the NATO member's land. Romanian officials had earlier denied reports of drones falling on Romanian territory and said Russian attacks in neighbouring Ukraine did not cause a direct threat. Tilvar told local news channel Antena 3 CNN on Wednesday that parts of what was most likely a drone were discovered in the eastern Tulcea county, an area of the Danube that forms a natural border between Romania and war-torn Ukraine. "I confirm that in this area, pieces that may be of a drone were found," he said, adding that the pieces did not pose a threat. He said the area had not been evacuated because there was nothing to suggest that the parts were dangerous and said the pieces would be analysed to confirm their origin.

The advent of generative AI, which includes chatbots such as OpenAI's ChatGPT and Google's Bard, has triggered concern that the technology could replace jobs, leading governments around the world to scramble to understand AI tools and respond. Prominent AI companies say they welcome regulation but have also lobbied against some approaches, saying strict laws could stifle the tech's development. There are also signs that consumer usage of generative AI tools is slowing, raising questions of how long the boom will last.