Space robotics poses unique challenges arising from the limitation of energy and computational resources, and the complexity of the environment and employed platforms. At the control center, offline motion planning is fundamental in the computation of optimized trajectories accounting for the system's constraints. Smooth movements, collision and forbidden areas avoidance, target visibility and energy consumption are all important factors to consider to be able to generate feasible and optimal plans. When mobile manipulators (terrestrial, aerial) are employed, the base and the arm movements are often separately planned, ultimately resulting in sub-optimal solutions. We propose an Optimal Whole Body Planner (OptiWB) based on Discrete Dynamic Programming (DDP) and optimal interpolation. Kinematic redundancy is exploited for collision and forbidden areas avoidance, and to improve target illumination and visibility from onboard cameras. The planner, implemented in ROS (Robot Operating System), interfaces 3DROCS, a mission planner used in several programs of the European Space Agency (ESA) to support planetary exploration surface missions and part of the ExoMars Rover's planning software. The proposed approach is exercised on a simplified version of the Analog-1 Interact rover by ESA, a 7-DOFs robotic arm mounted on a four wheels non-holonomic platform.

Abstract--With the rising popularity of Large Language Models (LLMs), assessing their trustworthiness through security tasks has gained critical importance. Regarding the new task of universal goal hijacking, previous efforts have concentrated solely on optimization algorithms, overlooking the crucial role of the prompt. To fill this gap, we propose a universal goal hijacking method called POUGH that incorporates semantic-guided prompt processing strategies. Specifically, the method starts with a sampling strategy to select representative prompts from a candidate pool, followed by a ranking strategy that prioritizes the prompts. Once the prompts are organized sequentially, the method employs an iterative optimization algorithm to generate the universal fixed suffix for the prompts. Experiments conducted on four popular LLMs and ten types of target responses verified the effectiveness of our method.

Multi-dimensional parameter spaces are commonly encountered in astroparticle physics theories that attempt to capture novel phenomena. However, they often possess complicated posterior geometries that are expensive to traverse using techniques traditional to this community. Effectively sampling these spaces is crucial to bridge the gap between experiment and theory. Several recent innovations, which are only beginning to make their way into this field, have made navigating such complex posteriors possible. These include GPU acceleration, automatic differentiation, and neural-network-guided reparameterization. We apply these advancements to astroparticle physics experimental results in the context of novel neutrino physics and benchmark their performances against traditional nested sampling techniques. Compared to nested sampling alone, we find that these techniques increase performance for both nested sampling and Hamiltonian Monte Carlo, accelerating inference by factors of $\sim 100$ and $\sim 60$, respectively. As nested sampling also evaluates the Bayesian evidence, these advancements can be exploited to improve model comparison performance while retaining compatibility with existing implementations that are widely used in the natural sciences.

Concerns about the effect of greenhouse gases have motivated the development of certification protocols to quantify the industrial carbon footprint (CF). These protocols are manual, work-intensive, and expensive. All of the above have led to a shift towards automatic data-driven approaches to estimate the CF, including Machine Learning (ML) solutions. Unfortunately, the decision-making processes involved in these solutions lack transparency from the end user's point of view, who must blindly trust their outcomes compared to intelligible traditional manual approaches. In this research, manual and automatic methodologies for CF estimation were reviewed, taking into account their transparency limitations. This analysis led to the proposal of a new explainable ML solution for automatic CF calculations through bank transaction classification. Consideration should be given to the fact that no previous research has considered the explainability of bank transaction classification for this purpose. For classification, different ML models have been employed based on their promising performance in the literature, such as Support Vector Machine, Random Forest, and Recursive Neural Networks. The results obtained were in the 90 % range for accuracy, precision, and recall evaluation metrics. From their decision paths, the proposed solution estimates the CO2 emissions associated with bank transactions. The explainability methodology is based on an agnostic evaluation of the influence of the input terms extracted from the descriptions of transactions using locally interpretable models. The explainability terms were automatically validated using a similarity metric over the descriptions of the target categories. Conclusively, the explanation performance is satisfactory in terms of the proximity of the explanations to the associated activity sector descriptions.

So far, the AI giants have largely played ball on paper. At the world's first AI Safety Summit six months ago, a bunch of tech bosses signed a voluntary pledge to create responsible, safe products that would maximise the benefits of AI technology and minimise its risks. Those risks, originally identified by the event organisers, were the proper stuff of nightmares. Six months later, when the summit reconvened, the word "safety" had been removed entirely from the conference title. Last week, a draft UK government report from a group of 30 independent experts concluded that there was "no evidence yet" that AI could generate a biological weapon or carry out a sophisticated cyber attack.

As voters will head to polls this year in more than 50 countries, experts have raised the alarm over AI-generated political disinformation and the prospect that malicious actors will use generative AI and social media to interfere with elections. Meta has previously faced criticism over its content moderation policies around past elections--for example, when it failed to prevent the January 6 rioters from organizing on its platforms. Clegg defended the company's efforts at preventing violent groups from organizing, but he also stressed the difficulty of keeping up. "This is a highly adversarial space. You remove one group, they rename themselves, rebrand themselves, and so on," he said.

Why not try a conversation with the leader of China? There's a new chatbot in town and it's based on Xi Jinping. As a matter of fact, it was trained using the'thoughts' of the Chinese leader. I put thoughts in quotes because researchers didn't use some kind of new mind-reading technology. Chinese officials just used a bunch of his books and papers for training purposes, according to a report by The Financial Times.

AI projects like OpenAI's ChatGPT get part of their savvy from some of the lowest-paid workers in the tech industry--contractors often in poor countries paid small sums to correct chatbots and label images. On Wednesday, 97 African workers who do AI training work or online content moderation for companies like Meta and OpenAI published an open letter to President Biden, demanding that US tech companies stop "systemically abusing and exploiting African workers." Most of the letter's signatories are from Kenya, a hub for tech outsourcing, whose president, William Ruto, is visiting the US this week. The workers allege that the practices of companies like Meta, OpenAI, and data provider Scale AI "amount to modern day slavery." The companies did not immediately respond to a request for comment.

On 21 May 2024, the Council of the European Union formally approved the artificial intelligence (AI) Act. The legislative act will come into force in about three weeks' time, with the new regulations being phased in over the course of the coming months and years. According to the Council, the new law aims to "foster the development and uptake of safe and trustworthy AI systems across the EU's single market by both private and public actors. At the same time, it aims to ensure respect of fundamental rights of EU citizens and stimulate investment and innovation on artificial intelligence in Europe." The legislation is designed to follow a risk-based approach, with the higher the risk a system poses, the stricter the rules relating to its use and/or release.

Sixteen companies at the forefront of developing artificial intelligence (AI) pledged on Tuesday at a global meeting to develop the technology safely at a time when regulators are scrambling to keep up with rapid innovation and emerging risks. The companies included U.S. leaders Google, Meta, Microsoft and OpenAI, as well as firms from China, South Korea and the United Arab Emirates. They were backed by a broader declaration from the Group of Seven (G7) major economies, the EU, Singapore, Australia and South Korea at a virtual meeting hosted by British Prime Minister Rishi Sunak and South Korean President Yoon Suk-yeol.