Visitors from far and wide have been traveling to Japan, with the country breaking a tourism record in 2024. Between Jan. 1 and Nov. 30, projections indicated that nearly 33.4 million travelers visited Japan, according to the country's government site. Nearly three million Americans visited the country in 2024. Hokuto Asano, first secretary at the Embassy of Japan, told Fox News Digital that the number of visitors last year ended up reaching 36 million. Yukiyoshi Noguchi, who is the counselor at the embassy, said 2024 was declared the "U.S.-Japan Tourism Year" by both governments.

Mass customization and shorter manufacturing cycles are becoming more important among small and medium-sized companies. However, classical industrial robots struggle to cope with product variation and dynamic environments. In this paper, we present CoBT, a collaborative programming by demonstration framework for generating reactive and modular behavior trees. CoBT relies on a single demonstration and a combination of data-driven machine learning methods with logic-based declarative learning to learn a task, thus eliminating the need for programming expertise or long development times. The proposed framework is experimentally validated on 7 manipulation tasks and we show that CoBT achieves approx. 93% success rate overall with an average of 7.5s programming time. We conduct a pilot study with non-expert users to provide feedback regarding the usability of CoBT.

Prompt injection has emerged as a serious security threat to large language models (LLMs). At present, the current best-practice for defending against newly-discovered prompt injection techniques is to add additional guardrails to the system (e.g., by updating the system prompt or using classifiers on the input and/or output of the model.) However, in the same way that variants of a piece of malware are created to evade anti-virus software, variants of a prompt injection can be created to evade the LLM's guardrails. Ideally, when a new prompt injection technique is discovered, candidate defenses should be tested not only against the successful prompt injection, but also against possible variants. In this work, we present, a tool to assist defenders in performing automated variant analysis of known prompt injection attacks. This involves solving two main challenges: (1) automatically generating variants of a given prompt according, and (2) automatically determining whether a variant was effective based only on the output of the model. This tool can also assist in generating datasets for jailbreak and prompt injection attacks, thus overcoming the scarcity of data in this domain. We evaluate Maatphor on three different types of prompt injection tasks. Starting from an ineffective (0%) seed prompt, Maatphor consistently generates variants that are at least 60% effective within the first 40 iterations.

Recently, there has been an increase in interest in evaluating large language models for emergent and dangerous capabilities. Importantly, agents could reason that in some scenarios their goal is better achieved if they are not turned off, which can lead to undesirable behaviors. In this paper, we investigate the potential of using toy textual scenarios to evaluate instrumental reasoning and shutdown avoidance in language models such as GPT-4 and Claude. Furthermore, we explore whether shutdown avoidance is merely a result of simple pattern matching between the dataset and the prompt or if it is a consistent behaviour across different environments and variations. We evaluated behaviours manually and also experimented with using language models for automatic evaluations, and these evaluations demonstrate that simple pattern matching is likely not the sole contributing factor for shutdown avoidance. This study provides insights into the behaviour of language models in shutdown avoidance scenarios and inspires further research on the use of textual scenarios for evaluations.

The dominant framework for off-policy multi-goal reinforcement learning involves estimating goal conditioned Q-value function. When learning to achieve multiple goals, data efficiency is intimately connected with the generalization of the Q-function to new goals. The de-facto paradigm is to approximate Q(s, a, g) using monolithic neural networks. To improve the generalization of the Q-function, we propose a bilinear decomposition that represents the Q-value via a low-rank approximation in the form of a dot product between two vector fields. The first vector field, f(s, a), captures the environment's local dynamics at the state s; whereas the second component, {\phi}(s, g), captures the global relationship between the current state and the goal. We show that our bilinear decomposition scheme substantially improves data efficiency, and has superior transfer to out-of-distribution goals compared to prior methods. Empirical evidence is provided on the simulated Fetch robot task-suite and dexterous manipulation with a Shadow hand.

Doing 500 days of AI project was a fascinating journey and enriched my life in many ways. One way was through awareness of the breadth of areas that artificial intelligence was being discussed within society. I could also more clearly see the varied applications of AI in multiple environments. After 500 days I looked back before Christmas in 2020 and I could say that I had at least the intention to get an understanding of the field of artificial intelligence. Here is a link to my article containing links to all the 500 articles on the topic of artificial intelligence.

The objective of transfer reinforcement learning is to generalize from a set of previous tasks to unseen new tasks. In this work, we focus on the transfer scenario where the dynamics among tasks are the same, but their goals differ. Although general value function (Sutton et al., 2011) has been shown to be useful for knowledge transfer, learning a universal value function can be challenging in practice. To attack this, we propose (1) to use universal successor representations (USR) to represent the transferable knowledge and (2) a USR approximator (USRA) that can be trained by interacting with the environment. Our experiments show that USR can be effectively applied to new tasks, and the agent initialized by the trained USRA can achieve the goal considerably faster than random initialization.

Few sci-fi tropes are more reliable in enthralling audiences than the plot of artificial intelligence betraying mankind. Perhaps this is because AI makes us confront the idea of what makes us human at all. From HAL 9000, to Skynet, to Westworld's robot uprising, the fears of sentient AI feel very real. Even Elon Musk worries about what AI is capable of. But are these fears unfounded?

U.S. Navy research teams recently completed combat trials with the branch's Tactical Battle Manager system using unmanned aerial vehicles. The Tactical Battle Manager system, or TBM, is a software platform designed to coordinate combat missions using "wingman" UAVs to assist manned and unmanned teams in combat. Researchers tested the system in a simulated beyond-visual-range combat scenario. The U.S. Naval Research Laboratory collaborated with the Naval Air Systems Command, the Navy Center for Applied Research in Artificial Intelligence and the Air Force Research Laboratory for the trials. During the tests, operators controlled a lead air vehicle and communicated with autonomous agents controlled by the TBM.

Today, Microsoft announced an audacious new goal: its researchers will attempt to "solve" cancer by treating the disease group as information processing systems that can be modeled and reasoned, and then use sophisticated analysis tools to better understand and treat cancer. "At Microsoft's research labs around the world, computer scientists, programmers, engineers and other experts are trying to use computer science to solve one of the most complex and deadly challenges humans face: Cancer." Microsoft's Allison Linn writes in a new post to Microsoft Stories. "And, for the most part, they are doing so with algorithms and computers instead of test tubes and beakers." Yes, Microsoft's research labs are hard at work on tough computer science problems too.