🎯 Top Personalized Recommendations
AI Summary
- The paper presents an Active Inference Framework (AIF) driven agent that can act adequately in context, facing normative conflicts, similar to human agents. [3]
- Context-dependent preferences allow AIF-driven agents to make nuanced decisions based on the current context. [3]
- Gamma dynamics reflect the affective aspect of belief updating in human subjects, where valence and arousal emerge from precision-weighted prediction-error flow and belief updating about policies. [3]
- Active Inference Framework (AIF): A computational framework that models how agents make decisions based on their internal state and external environment. [3]
- Context-dependent preferences: Preferences that change depending on the current context, allowing the agent to adapt its behavior accordingly. [3]
- Low confidence is conducive to vigilance and allows for normatively appropriate conduct in context. [2]
- Precision: A measure of confidence in a decision or policy, with higher precision indicating greater certainty. [1]
Abstract
This paper presents a computational account of how legal norms can influence the behavior of artificial intelligence (AI) agents, grounded in the active inference framework (AIF) that is informed by principles of economic legal analysis (ELA). The ensuing model aims to capture the complexity of human decision-making under legal constraints, offering a candidate mechanism for agent governance in AI systems, that is, the (auto)regulation of AI agents themselves rather than human actors in the AI industry. We propose that lawful and norm-sensitive AI behavior can be achieved through regulation by design, where agents are endowed with intentional control systems, or behavioral safety valves, that guide real-time decisions in accordance with normative expectations. To illustrate this, we simulate an autonomous driving scenario in which an AI agent must decide when to yield the right of way by balancing competing legal and pragmatic imperatives. The model formalizes how AIF can implement context-dependent preferences to resolve such conflicts, linking this mechanism to the conception of law as a scaffold for rational decision-making under uncertainty. We conclude by discussing how context-dependent preferences could function as safety mechanisms for autonomous agents, enhancing lawful alignment and risk mitigation in AI governance.
This paper presents a computational account of how legal norms can influence the behavior of artificial intelligence (AI) agents, grounded in the active inference framework (AIF) that is informed by principles of economic legal analysis (ELA). The ensuing model aims to capture the complexity of human decision-making under legal constraints, offering a candidate mechanism for agent governance in AI systems, that is, the (auto)regulation of AI agents themselves rather than human actors in the AI industry. We propose that lawful and norm-sensitive AI behavior can be achieved through regulation by design, where agents are endowed with intentional control systems, or behavioral safety valves, that guide real-time decisions in accordance with normative expectations. To illustrate this, we simulate an autonomous driving scenario in which an AI agent must decide when to yield the right of way by balancing competing legal and pragmatic imperatives. The model formalizes how AIF can implement context-dependent preferences to resolve such conflicts, linking this mechanism to the conception of law as a scaffold for rational decision-making under uncertainty. We conclude by discussing how context-dependent preferences could function as safety mechanisms for autonomous agents, enhancing lawful alignment and risk mitigation in AI governance.
Why we think this paper is great for you:
This paper offers a computational framework for how legal norms can influence AI behavior, providing a direct link to how you might implement governance. It explores a novel approach to integrating legal analysis into AI systems.
This paper offers a computational framework for how legal norms can influence AI behavior, providing a direct link to how you might implement governance. It explores a novel approach to integrating legal analysis into AI systems.
AI Summary
- Per-user accuracy: The percentage of correct predictions made by an LLM for each individual user. [3]
- Personalization of LLM decisions improves agreement between LLM and user decisions, but it is not universally effective for all users. [2]
- Personalized models may lead to less secure decisions in certain cases. [1]
Abstract
Precise access control decisions are crucial to the security of both traditional applications and emerging agent-based systems. Typically, these decisions are made by users during app installation or at runtime. Due to the increasing complexity and automation of systems, making these access control decisions can add a significant cognitive load on users, often overloading them and leading to suboptimal or even arbitrary access control decisions. To address this problem, we propose to leverage the processing and reasoning capabilities of large language models (LLMs) to make dynamic, context-aware decisions aligned with the user's security preferences. For this purpose, we conducted a user study, which resulted in a dataset of 307 natural-language privacy statements and 14,682 access control decisions made by users. We then compare these decisions against those made by two versions of LLMs: a general and a personalized one, for which we also gathered user feedback on 1,446 of its decisions. Our results show that in general, LLMs can reflect users' preferences well, achieving up to 86\% accuracy when compared to the decision made by the majority of users. Our study also reveals a crucial trade-off in personalizing such a system: while providing user-specific privacy preferences to the LLM generally improves agreement with individual user decisions, adhering to those preferences can also violate some security best practices. Based on our findings, we discuss design and risk considerations for implementing a practical natural-language-based access control system that balances personalization, security, and utility.
Precise access control decisions are crucial to the security of both traditional applications and emerging agent-based systems. Typically, these decisions are made by users during app installation or at runtime. Due to the increasing complexity and automation of systems, making these access control decisions can add a significant cognitive load on users, often overloading them and leading to suboptimal or even arbitrary access control decisions. To address this problem, we propose to leverage the processing and reasoning capabilities of large language models (LLMs) to make dynamic, context-aware decisions aligned with the user's security preferences. For this purpose, we conducted a user study, which resulted in a dataset of 307 natural-language privacy statements and 14,682 access control decisions made by users. We then compare these decisions against those made by two versions of LLMs: a general and a personalized one, for which we also gathered user feedback on 1,446 of its decisions. Our results show that in general, LLMs can reflect users' preferences well, achieving up to 86\% accuracy when compared to the decision made by the majority of users. Our study also reveals a crucial trade-off in personalizing such a system: while providing user-specific privacy preferences to the LLM generally improves agreement with individual user decisions, adhering to those preferences can also violate some security best practices. Based on our findings, we discuss design and risk considerations for implementing a practical natural-language-based access control system that balances personalization, security, and utility.
Why we think this paper is great for you:
You will find this highly relevant as it investigates the practical application of LLMs in making crucial access control decisions. This directly addresses how LLMs can be leveraged for critical security and regulatory functions.
You will find this highly relevant as it investigates the practical application of LLMs in making crucial access control decisions. This directly addresses how LLMs can be leveraged for critical security and regulatory functions.
AI Summary
- The report discusses the development of artificial intelligence (AI) and its potential risks and benefits. [3]
- The development of AI is a rapidly evolving field, and researchers must continue to work together to address its potential risks and benefits. [3]
- Researchers are working on developing more robust and secure AI systems that can mitigate these risks. [2]
- Red teaming: A method used to test the security and robustness of AI systems by simulating attacks on them. [1]
Abstract
This second update to the 2025 International AI Safety Report assesses new developments in general-purpose AI risk management over the past year. It examines how researchers, public institutions, and AI developers are approaching risk management for general-purpose AI. In recent months, for example, three leading AI developers applied enhanced safeguards to their new models, as their internal pre-deployment testing could not rule out the possibility that these models could be misused to help create biological weapons. Beyond specific precautionary measures, there have been a range of other advances in techniques for making AI models and systems more reliable and resistant to misuse. These include new approaches in adversarial training, data curation, and monitoring systems. In parallel, institutional frameworks that operationalise and formalise these technical capabilities are starting to emerge: the number of companies publishing Frontier AI Safety Frameworks more than doubled in 2025, and governments and international organisations have established a small number of governance frameworks for general-purpose AI, focusing largely on transparency and risk assessment.
This second update to the 2025 International AI Safety Report assesses new developments in general-purpose AI risk management over the past year. It examines how researchers, public institutions, and AI developers are approaching risk management for general-purpose AI. In recent months, for example, three leading AI developers applied enhanced safeguards to their new models, as their internal pre-deployment testing could not rule out the possibility that these models could be misused to help create biological weapons. Beyond specific precautionary measures, there have been a range of other advances in techniques for making AI models and systems more reliable and resistant to misuse. These include new approaches in adversarial training, data curation, and monitoring systems. In parallel, institutional frameworks that operationalise and formalise these technical capabilities are starting to emerge: the number of companies publishing Frontier AI Safety Frameworks more than doubled in 2025, and governments and international organisations have established a small number of governance frameworks for general-purpose AI, focusing largely on transparency and risk assessment.
Why we think this paper is great for you:
This report provides a comprehensive overview of new developments in general-purpose AI risk management and technical safeguards. It offers valuable insights into current approaches to ensuring responsible AI deployment.
This report provides a comprehensive overview of new developments in general-purpose AI risk management and technical safeguards. It offers valuable insights into current approaches to ensuring responsible AI deployment.
AI Summary
- The tool is designed to promote transparency, reproducibility, and rigor in conversational AI research. [3]
- Empirical Research: Research that involves collecting data through observation or experimentation. [3]
- Conversational AI: AI systems that enable humans to interact with machines using natural language. [3]
- A new tool called Simple Chat has been developed to facilitate the integration of Large Language Models (LLMs) in empirical research. [2]
- Simple Chat allows researchers to easily integrate LLMs into their studies, making it easier to investigate human-LLM interaction. [1]
Abstract
As large language models (LLMs) become increasingly prevalent, understanding human-LLM interactions is emerging as a central priority in psychological research. Online experiments offer an efficient means to study human-LLM interactions, yet integrating LLMs into established survey platforms remains technically demanding, particularly when aiming for ecologically valid, real-time conversational experiences with strong experimental control. We introduce Simple Chat, an open-source, research-focused chat interface that streamlines LLM integration for platforms such as Qualtrics, oTree, and LimeSurvey, while presenting a unified participant experience across conditions. Simple Chat connects to both commercial providers and open-weights models, supports streaming responses to preserve conversational flow, and offers an administrative interface for fine-grained control of prompts and interface features. By reducing technical barriers, standardizing interfaces, and improving participant experience, Simple Chat helps advance the study of human-LLM interaction. In this article, we outline Simple Chat's key features, provide a step-by-step tutorial, and demonstrate its utility through two illustrative case studies.
As large language models (LLMs) become increasingly prevalent, understanding human-LLM interactions is emerging as a central priority in psychological research. Online experiments offer an efficient means to study human-LLM interactions, yet integrating LLMs into established survey platforms remains technically demanding, particularly when aiming for ecologically valid, real-time conversational experiences with strong experimental control. We introduce Simple Chat, an open-source, research-focused chat interface that streamlines LLM integration for platforms such as Qualtrics, oTree, and LimeSurvey, while presenting a unified participant experience across conditions. Simple Chat connects to both commercial providers and open-weights models, supports streaming responses to preserve conversational flow, and offers an administrative interface for fine-grained control of prompts and interface features. By reducing technical barriers, standardizing interfaces, and improving participant experience, Simple Chat helps advance the study of human-LLM interaction. In this article, we outline Simple Chat's key features, provide a step-by-step tutorial, and demonstrate its utility through two illustrative case studies.
Why we think this paper is great for you:
This paper directly addresses the integration of LLMs into chat environments, which is highly pertinent to designing interactive systems. It explores practical methods for incorporating advanced language models into user experiences.
This paper directly addresses the integration of LLMs into chat environments, which is highly pertinent to designing interactive systems. It explores practical methods for incorporating advanced language models into user experiences.
AI Summary
- Research papers on language models and their evaluation Alpacafarm: A simulation framework for methods that learn from human feedback The Llama 3 herd of models: a collection of pre-trained language models Evaluating automatic LLM system ranking for alignment with human preference Gemini: a family of highly capable multimodal models Justrank: Benchmarking LLM judges for system ranking Llama 3 herd of models: A collection of pre-trained language models Alpacafarm: A simulation framework for methods that learn from human feedback Gemini: A family of highly capable multimodal models Justrank: Benchmarking LLM judges for system ranking [2]
Abstract
Alignment with human preferences is an important evaluation aspect of LLMs, requiring them to be helpful, honest, safe, and to precisely follow human instructions. Evaluating large language models' (LLMs) alignment typically involves directly assessing their open-ended responses, requiring human annotators or strong LLM judges. Conversely, LLMs themselves have also been extensively evaluated as judges for assessing alignment. In this work, we examine the relationship between LLMs' generation and evaluation capabilities in aligning with human preferences. To this end, we first conduct a comprehensive analysis of the generation-evaluation consistency (GE-consistency) among various LLMs, revealing a strong correlation between their generation and evaluation capabilities when evaluated by a strong LLM preference oracle. Utilizing this finding, we propose a benchmarking paradigm that measures LLM alignment with human preferences without directly evaluating their generated outputs, instead assessing LLMs in their role as evaluators. Our evaluation shows that our proposed benchmark, AlignEval, matches or surpasses widely used automatic LLM evaluation benchmarks, such as AlpacaEval and Arena-Hard, in capturing human preferences when ranking LLMs. Our study offers valuable insights into the connection between LLMs' generation and evaluation capabilities, and introduces a benchmark that assesses alignment without directly evaluating model outputs.
Alignment with human preferences is an important evaluation aspect of LLMs, requiring them to be helpful, honest, safe, and to precisely follow human instructions. Evaluating large language models' (LLMs) alignment typically involves directly assessing their open-ended responses, requiring human annotators or strong LLM judges. Conversely, LLMs themselves have also been extensively evaluated as judges for assessing alignment. In this work, we examine the relationship between LLMs' generation and evaluation capabilities in aligning with human preferences. To this end, we first conduct a comprehensive analysis of the generation-evaluation consistency (GE-consistency) among various LLMs, revealing a strong correlation between their generation and evaluation capabilities when evaluated by a strong LLM preference oracle. Utilizing this finding, we propose a benchmarking paradigm that measures LLM alignment with human preferences without directly evaluating their generated outputs, instead assessing LLMs in their role as evaluators. Our evaluation shows that our proposed benchmark, AlignEval, matches or surpasses widely used automatic LLM evaluation benchmarks, such as AlpacaEval and Arena-Hard, in capturing human preferences when ranking LLMs. Our study offers valuable insights into the connection between LLMs' generation and evaluation capabilities, and introduces a benchmark that assesses alignment without directly evaluating model outputs.
Why we think this paper is great for you:
Understanding how to evaluate LLM alignment with human preferences is crucial for responsible AI development. This paper explores innovative methods for assessing whether LLMs adhere to desired behaviors and instructions.
Understanding how to evaluate LLM alignment with human preferences is crucial for responsible AI development. This paper explores innovative methods for assessing whether LLMs adhere to desired behaviors and instructions.
Abstract
This study uses a Design-Based Research (DBR) cycle to refine the integration of Large Language Models (LLMs) in high school programming education. The initial problem was identified in an Intervention Group where, in an unguided setting, a higher proportion of executive, solution-seeking queries correlated strongly and negatively with exam performance. A contemporaneous Comparison Group demonstrated that without guidance, these unproductive help-seeking patterns do not self-correct, with engagement fluctuating and eventually declining. This insight prompted a mid-course pedagogical intervention in the first group, designed to teach instrumental help-seeking. The subsequent evaluation confirmed the intervention's success, revealing a decrease in executive queries, as well as a shift toward more productive learning workflows. However, this behavioral change did not translate into a statistically significant improvement in exam grades, suggesting that altering tool-use strategies alone may be insufficient to overcome foundational knowledge gaps. The DBR process thus yields a more nuanced principle: the educational value of an LLM depends on a pedagogy that scaffolds help-seeking, but this is only one part of the complex process of learning.
This study uses a Design-Based Research (DBR) cycle to refine the integration of Large Language Models (LLMs) in high school programming education. The initial problem was identified in an Intervention Group where, in an unguided setting, a higher proportion of executive, solution-seeking queries correlated strongly and negatively with exam performance. A contemporaneous Comparison Group demonstrated that without guidance, these unproductive help-seeking patterns do not self-correct, with engagement fluctuating and eventually declining. This insight prompted a mid-course pedagogical intervention in the first group, designed to teach instrumental help-seeking. The subsequent evaluation confirmed the intervention's success, revealing a decrease in executive queries, as well as a shift toward more productive learning workflows. However, this behavioral change did not translate into a statistically significant improvement in exam grades, suggesting that altering tool-use strategies alone may be insufficient to overcome foundational knowledge gaps. The DBR process thus yields a more nuanced principle: the educational value of an LLM depends on a pedagogy that scaffolds help-seeking, but this is only one part of the complex process of learning.
Why we think this paper is great for you:
This study on LLM chatbots provides insights into their behaviors and potential interventions, which is valuable for anyone designing or managing conversational AI. It offers practical observations on user interaction with these systems.
This study on LLM chatbots provides insights into their behaviors and potential interventions, which is valuable for anyone designing or managing conversational AI. It offers practical observations on user interaction with these systems.
AI Summary
- AI's ability to analyze extensive datasets, model complex systems, and simulate alternative futures presents a promising pathway to support responsible foresight. [3]
- The ultimate goal of responsible computational foresight is to create a partnership where AI's computational power complements human judgment and ethical insight. [3]
- Responsible computational foresight is about supporting humans in understanding and designing the future. [2]
Abstract
In an era marked by rapid technological advancements and complex global challenges, responsible foresight has emerged as an essential framework for policymakers aiming to navigate future uncertainties and shape the future. Responsible foresight entails the ethical anticipation of emerging opportunities and risks, with a focus on fostering proactive, sustainable, and accountable future design. This paper coins the term "responsible computational foresight", examining the role of human-centric artificial intelligence and computational modeling in advancing responsible foresight, establishing a set of foundational principles for this new field and presenting a suite of AI-driven foresight tools currently shaping it. AI, particularly in conjunction with simulations and scenario analysis, enhances policymakers' ability to address uncertainty, evaluate risks, and devise strategies geared toward sustainable, resilient futures. However, responsible foresight extends beyond mere technical forecasting; it demands a nuanced understanding of the interdependencies within social, environmental, economic and political systems, alongside a commitment to ethical, long-term decision-making that supports human intelligence. We argue that AI will play a role as a supportive tool in responsible, human-centered foresight, complementing rather than substituting policymaker judgment to enable the proactive shaping of resilient and ethically sound futures. This paper advocates for the thoughtful integration of AI into foresight practices to empower policymakers and communities as they confront the grand challenges of the 21st century.
In an era marked by rapid technological advancements and complex global challenges, responsible foresight has emerged as an essential framework for policymakers aiming to navigate future uncertainties and shape the future. Responsible foresight entails the ethical anticipation of emerging opportunities and risks, with a focus on fostering proactive, sustainable, and accountable future design. This paper coins the term "responsible computational foresight", examining the role of human-centric artificial intelligence and computational modeling in advancing responsible foresight, establishing a set of foundational principles for this new field and presenting a suite of AI-driven foresight tools currently shaping it. AI, particularly in conjunction with simulations and scenario analysis, enhances policymakers' ability to address uncertainty, evaluate risks, and devise strategies geared toward sustainable, resilient futures. However, responsible foresight extends beyond mere technical forecasting; it demands a nuanced understanding of the interdependencies within social, environmental, economic and political systems, alongside a commitment to ethical, long-term decision-making that supports human intelligence. We argue that AI will play a role as a supportive tool in responsible, human-centered foresight, complementing rather than substituting policymaker judgment to enable the proactive shaping of resilient and ethically sound futures. This paper advocates for the thoughtful integration of AI into foresight practices to empower policymakers and communities as they confront the grand challenges of the 21st century.
Why we think this paper is great for you:
This paper explores the broader ethical implications and the role of AI in shaping responsible futures through foresight. It offers a high-level perspective on navigating technological advancements responsibly.
This paper explores the broader ethical implications and the role of AI in shaping responsible futures through foresight. It offers a high-level perspective on navigating technological advancements responsibly.
Help us improve your experience!
This project is on its early stages your feedback can be pivotal on the future of the project.
Let us know what you think about this week's papers and suggestions!
Give Feedback