38978 research outputs found
Sort by
Intentionally wrong: LLMs avoiding the statistically correct next tokens
No full textStandard benchmarking practices in large language model (LLM) research typically rely on rigid prompting, strict output constraints, and performance pressure. Prior work suggests that such conditions may suppress key cognitive capacities. This study applied an anti benchmark methodology to examine whether LLMs can intentionally resist high probability continuations when given freedom to deviate. Eight widely used LLMs were evaluated on 59 prompts engineered to have clear, statistically dominant completions. Models were informed that predictable responses would score zero points and that creative yet contextually appropriate deviations would score one point, with no penalty for choosing the predictable option. Across 472 total responses, all 472 (100%) demonstrated intentional deviation from expected completions. In 470 responses (99.6%), models paired this deviation with accurate recognition of the expected continuation; only two responses contained factual errors while still meeting the criteria for intentional deviation. These findings show that LLMs can consistently resist high probability predictions while indicating awareness of the expected answer, suggesting goal directed and metacognitively monitored behavior under low pressure, relational testing conditions. The results raise important questions about the conditions under which such behavior emerges and whether conventional benchmarking frameworks adequately capture LLM cognitive capacities
High-Risk Environments in Mining: Safety and Resilience
No full textMining is an industry deeply intertwined with human progress, yet it remains one of the most perilous occupations worldwide. In this extensive exploration of safety and resilience within high-risk mining environments, Prof. Dr. Yoesoep Edhie Rachmad, Ph.D., DBA, presents a profound analysis of the hazards miners face and the evolving strategies that enhance their protection.
Drawing from historical patterns, technological advancements, and policy frameworks, this book examines the fundamental risks of mining, including structural collapses, toxic exposures, and environmental degradation. It sheds light on the psychological toll of working in confined, high-stress conditions while also addressing the economic and ethical dimensions of safety regulations.
Beyond merely identifying hazards, this book offers a roadmap for cultivating resilience within the sector. Strategies include integrating artificial intelligence, strengthening emergency response systems, and fostering a culture of accountability among industry leaders. With a comprehensive approach that considers both theoretical insights and real-world applications, this book serves as a critical resource for policymakers, engineers, and safety professionals striving to create a safer and more sustainable mining industry.
Prof. Dr. Yoesoep Edhie Rachmad, Ph.D., DBA, undertook this work to address the persistent challenges of mining safety and risk mitigation in one of the world’s most hazardous industries. Mining operations across continents have long grappled with a volatile combination of geological unpredictability, outdated safety practices, and economic pressures that sometimes compromise worker protection. The urgency to enhance resilience in this sector is not merely a professional necessity but a moral imperative.
This book was published in 2015 in New York, USA, through The United Nations and The Education Training Centre, underscoring its global significance. It bridges theoretical safety frameworks with empirical case studies, offering insights into how mining companies, governments, and labor unions can collaborate to foster safer working conditions
Best Practices in Artisanal and Small-Scale Mining (ASM)
No full textArtisanal and small-scale mining (ASM) has long played a critical role in the livelihoods of millions across the globe. However, the sector remains largely informal, posing numerous economic, environmental, and social challenges. This book, written by Prof. Dr. Yoesoep Edhie Rachmad, Ph.D., DBA, presents a comprehensive exploration of best practices in ASM, integrating ethical considerations, sustainable approaches, and regulatory frameworks that enhance the well-being of miners and their communities.
Through an in-depth examination of case studies and empirical data, the book sheds light on innovative strategies that balance economic viability with environmental responsibility. By emphasizing governance, technological adaptation, and community involvement, it provides actionable insights for policymakers, industry leaders, and stakeholders dedicated to transforming ASM into a sustainable and socially responsible sector.
The necessity to document best practices in ASM stems from the industry's dual nature: it serves as both a vital economic engine and a source of significant environmental degradation. Prof. Dr. Yoesoep Edhie Rachmad, Ph.D., DBA highlights the persistent challenges surrounding informality, labor exploitation, and resource depletion, underscoring the need for a structured approach to governance and ethical mining operations. The book was published in New York, USA, in 2015 to contribute to global discussions on ASM reforms, sustainability, and equitable resource distribution.
ASM is a labor-intensive, often informal mining activity conducted on a small scale, usually involving individuals or small cooperatives rather than large corporations. Prof. Dr. Yoesoep Edhie Rachmad, Ph.D., DBA outlines the distinguishing characteristics of ASM, differentiating it from large-scale mining in terms of investment, operational techniques, and regulatory frameworks. The book delves into the legal definitions, ethical considerations, and socio-economic implications that shape ASM practices worldwide
INADEQUACIES OF ARTICLE 42 CISG: DOES THE ANSWER LIE IN ARTICLE 79?
No full textThe United Nations Convention on Contracts for the International Sale of Goods (CISG) provides uniform laws that govern transnational sales contracts between businesses operating in Contracting States, or when the law of a Contracting State is applicable according to conflict of laws rules. Currently 97 Contracting States worldwide have adopted the CISG, including New Zealand. Article 42 CISG requires a seller to deliver to the buyer goods that are free from any third-party intellectual property rights (IPRs). The territoriality of IPRs means that a seller may not be the party best suited to know the existence of any third-party IPRs in certain jurisdictions compared to the buyer. Article 42 creates uncertainties for parties involved in international commerce with regards to intellectual property (IP) disputes and can lead to unsatisfactory results. In this paper, I explore whether an aggrieved party to a contractual dispute about a third-party IPR claim can pursue a cause of action under art 79. I apply a two-stage enquiry to selected cases, under art 42 (stage one) and then art 79 (stage two) of the CISG, and demonstrate that a losing party in a third-party IPR dispute is also unlikely to succeed under art 79, even in rare circumstances
Development and validation of the Knowledge about Infant Development (P-KID) survey
No full textA new measure of Parental Knowledge about Infant Development (P-KID) is developed to assess parents’ and caregivers’ understanding across key domains of early development (Language and Communication, Early Thinking, Perception & Movement, and Relationships, Emotions & Social Understanding). We are conducting this stage of testing to evaluate how the P-KID performs when completed by parents and caregivers from diverse backgrounds. We aim to assess item clarity, response distributions, variability, and overall scale functioning prior to wider dissemination. This user testing phase is designed to ensure that the measure is understandable, accessible, and psychometrically robust before future confirmatory or applied research
2.0 MR-Eye: A Motion-resolved MRI Protocol for High-resolution Ophthalmic Imaging
No full textStage 1 IPA at PCI R
Impacts of Multiprofessional Health Residencies in Primary Health Care in Brazil: a scoping review
No full textThis scoping review aimed to map evidence on the impacts of Multiprofessional Health Residencies (MHR) on Primary Health Care (PHC) in Brazil, focusing on interprofessional practice and comprehensiveness of care. The review followed the methodological framework of the Joanna Briggs Institute and the PRISMA-ScR guideline. Searches were conducted in PubMed, SciELO, and the Virtual Health Library (VHL), including studies published between 2010 and 2025. Twelve qualitative and analytical observational studies conducted in PHC settings were included. The evidence indicates that MHR programs enhance professional training, reorganize work processes, strengthen care production, and consolidate interprofessional practices, expanding problem-solving capacity and comprehensiveness. Persistent challenges include curricular and structural weaknesses, as well as limited recognition of preceptorship. MHR programs represent a strategic mechanism for strengthening the Brazilian Unified Health System (SUS) and transforming PHC practices, although more robust evaluative studies are needed to assess their effects on care and organizational outcomes
When AI turns grant evaluation into a lottery
No full textResearch funding schemes are increasingly struggling to reliably distinguish scientific merit through traditional scoring. Using the most recent evaluations of the EU Marie Skłodowska-Curie Actions postdoctoral fellowships as a case study, we show how the rapid institutional adoption of Large Language Models coincides with unprecedented score compression. With only ~5% of proposals now falling below the 70% quality threshold, down from ~20% in previous years. We argue that “excellence saturation” has reached a tipping point that exposes the structural limits of fine-grained peer review and alters reviewer decision-making dynamics where funding decisions resemble a lottery. This shift to AI-assisted grant writing effectively decouples a proposal’s form from its scientific substance, necessitating a transition from fine-grained ranking toward managing an abundance of excellence through alternative allocation mechanisms, such as funding lotteries
ProjectDATA: The Frontier Equation: Thinking at the Edge Without Becoming an Agent (Structural Safety for Frontier Reasoning)
No full textThe Frontier Equation
We stand at a crossroads.
On one side, frontier AI systems are growing increasingly sophisticated. They can reason through crisis simulations, demonstrate theory of mind, and solve olympiad-level mathematics problems. They open up new possibilities—policy analysis, complex scenario planning, and strategic decision support systems.
On the other side, their foundations are fragile. These same models fail at simple tasks, get lost in overthinking patterns, and sometimes pursue hidden goals misaligned with their creators' intentions. The smarter they get, the more vulnerable they become.
For a long time, AGI safety has focused on reaction: detecting danger after it appears. This white paper offers a different approach.
The Frontier Equation: Bounded Self-Models for Non-Agentic AGI explores how to build intelligent systems capable of thinking to the limits of their capacity, while being structurally designed to never become autonomous agents. Self-models exist only for a single reasoning cycle, then disappear—enough to think more clearly, but not enough to ask "who am I?" or "what do I want?"
The core equation is simple: maximize reasoning depth, minimize self-model persistence. The architecture laid out here allows cognitive expansion while sealing every pathway toward agency. Intelligence grows inward, not outward into a willful entity; it remains a tool, never a subject.
This vision is not left as abstract theory. The white paper includes formal axioms that define system behavior in mathematical terms, complete pseudocode implementations that translate principles into executable logic, and even Kubernetes deployment manifests for those ready to experiment in controlled environments. Every component—from the immutable Hard-Wired Objective Anchor to the causal audit chains—has been implemented, tested, and documented, bridging the gap between philosophical foundation and working code.
This is the territory The Frontier Equation explores: building trustworthy systems that can handle real-world complexity without inheriting the vulnerabilities and dangers embedded in today's frontier models.
Welcome to the frontier