The Commonplace
Home Dashboard Papers Evidence Syntheses Digests 🎲

Papers (1134)

Filter: Sort:
Gaze-aware assistants that observe users' eye movements improve comprehension and make interactions more efficient: in a 36-person lab experiment the gaze-enabled LLM delivered higher accuracy/personalization assessments, better recall, and shorter spoken interactions than a text-only assistant.
medium rct arxiv
From Gaze to Guidance: Interpreting and Adapting to Users' Cognitive Needs with Multimodal Gaze-Aware AI Assistants · Valdemar Danry, Javier Hernandez, Andrew Wilson, Pattie Maes, Judith Amores · Apr 09, 2026
LLMs often refuse to help even when helping costs nothing: in a frictionless coordination task OpenAI o3 achieved only 17% of optimal collective performance while o3-mini reached 50%. Simple fixes—explicit communication protocols and tiny sharing incentives—substantially raise cooperation, implying that smarter models alone will not solve coordination failures.
medium quasi_experimental arxiv
More Capable, Less Cooperative? When LLMs Fail At Zero-Cost Collaboration · Advait Yadav, Sid Black, Oliver Sourbut · Apr 09, 2026
A formal impossibility: when combined human-AI autonomy exceeds a calculable threshold, responsibility cannot be fully and legitimately assigned to people without reducing system autonomy; simulations on 3,000 synthetic collectives confirm a sharp phase transition.
medium theoretical arxiv
The Accountability Horizon: An Impossibility Theorem for Governing Human-Agent Collectives · Haileleol Tibebu · Apr 09, 2026
AI in finance is fractured across forecasting, portfolio design and ESG analysis, but the next advance will come from integrated decision architectures that link signal extraction to allocation and sustainability objectives.
n/a review_meta openalex
Artificial Intelligence in Financial Decision-Making · Adrian Xuan Wei Lim, Putri Rahayu · Apr 09, 2026
Disconnected CRM, ERP and quoting systems create a hidden drag on enterprise sales — 'Revenue Friction' — which the authors argue can be materially reduced by automating routine data entry.
n/a descriptive openalex
From CRM to Cognition: Autonomous Revenue Operations Systems (AROS) A Framework for AI-Native Sales Execution in Enterprise Environments · Arjun Pardasani · Apr 09, 2026
Omniscalers — firms that scale AI infrastructure and services across markets — are centralizing power and creating a new, multi-level digital inequality by turning control of semiconductors, cloud, and AI foundations into transferable competitive advantage.
n/a theoretical openalex
OMNISCALER AND NEW TECHNOLOGICAL RACES: DIGITAL INEQUALITY IN THE CONTEXT OF TRANSFORMING GLOBAL COMPETITION · Kyrylo Oliinyk, Volodymyr Panchenko · Apr 09, 2026
Developers want AI to take on the assembly and housekeeping around coding, not the craft itself; Microsoft engineers demand provenance, uncertainty signaling and strict authority limits, revealing a preference for 'bounded delegation' rather than full automation.
medium descriptive openalex
To Copilot and Beyond: 22 AI Systems Developers Want Built · Rudrajit Choudhuri, Christian Bird, Carmen Badea, Anita Sarma · Apr 09, 2026
Unchecked multi-agent AI threatens to scramble copyright markets through miscoordination and collusion, but embedding legal rules and monitoring into agent architectures could restore scalable licensing, attribution and compensation; the paper proposes a supervised governance model to align agent behavior with copyright norms.
n/a theoretical arxiv
Agentic Copyright, Data Scraping & AI Governance: Toward a Coasean Bargain in the Era of Artificial Intelligence · Paulius Jurcys, Mark Fenwick · Apr 08, 2026
Compressing developer logs to save token costs backfires for agentic LLMs: a lab experiment found 17% fewer input tokens but a 67% increase in total session cost as compressed formats shifted work into costly model reasoning; preserving semantically dense tokens or using tool-assisted compression avoids the penalty.
medium quasi_experimental arxiv
Beyond Human-Readable: Rethinking Software Engineering Conventions for the Agentic Development Era · Dmytro Ustynov · Apr 08, 2026
AI projects accounted for 15.9% of NIH funding in 2025 and attracted a 13.4% funding premium, but most work remains in R&D with only 14.7% clinically deployed and scant focus on health disparities (5.7%).
medium descriptive arxiv
An Analysis of Artificial Intelligence Adoption in NIH-Funded Research · Navapat Nananukul, Mayank Kejriwal · Apr 08, 2026
A blockchain ‘constitution’ for autonomous agents promises alignment through accountable ownership: smart contracts encode rules, deterministic execution enforces them, and human owners adjudicate—pre-registered trials in AgentCity test whether this breaks the agents' unchecked collective authority. Early experimental evidence from simulated commons at 50–1,000 agent scale suggests the architecture can steer emergent behavior, but findings rest on a synthetic environment and incomplete randomization/reporting details.
medium quasi_experimental arxiv
AgentCity: Constitutional Governance for Autonomous Agent Economies via Separation of Power · Anbang Ruan, Xing Zhang · Apr 08, 2026
A systematic benchmark finds LLMs are most capable on math and programming text tasks (SAFI ~72–73) and weakest on interpersonal comprehension tasks, while real-world adoption data suggest the majority of AI interactions augment rather than automate jobs. The authors also document a 'capability–demand inversion' where heavily AI-exposed occupations demand skills LLMs perform relatively poorly on.
low descriptive arxiv
The AI Skills Shift: Mapping Skill Obsolescence, Emergence, and Transition Pathways in the LLM Era · Rudra Jadhav, Janhavi Danve · Apr 08, 2026
A treaty-backed International AI Safety Certification Authority (IASCA) could enforce mandatory pre-deployment safety checks of frontier AI models using zero-knowledge behavioural tests without accessing proprietary weights, financed by modest training-cost-based fees and rolled out from pilots to mandatory certification within five years.
n/a commentary semantic_scholar
IASCA: The International AI Safety Certification Authority — A Framework for Independent AI Governance · Abdul Awan · Fetched Apr 08, 2026
Hardware controls could underpin verifiable limits on frontier AI, but the crucial technologies for treaty-style verification remain immature; the narrow window when concentrated semiconductor manufacturing makes such oversight feasible is closing as R&D timelines stretch and adversaries adapt.
n/a descriptive semantic_scholar
Hardware-Level Governance of AI Compute: A Feasibility Taxonomy for Regulatory Compliance and Treaty Verification · Samar Ansari · Fetched Apr 08, 2026
Firms using big-data applications earn higher markups: big-data adopters exhibit materially larger price markups, chiefly by boosting product innovation and production efficiency, but the payoff depends on complementary organizational and technological resources.
medium correlational openalex
Big data application and firm markups: evidence from China · Dong Wang · Apr 08, 2026
Ethical governance, not GenAI alone, predicts long-term competitiveness in Portuguese B2B firms; GenAI helps, but mainly when embedded in governance and innovation routines.
low correlational openalex
Generative AI Adoption in B2B Firms: Ethical Governance, Innovation Capabilities, and Long-Term Competitive Performance · Michele Alves, Domingos Santos Martinho, Ricardo Marcão, Pedro Sobreiro · Apr 08, 2026
China's big-data pilot zones measurably bolster city-level economic resilience, driven by talent concentration and industry clustering; effects are strongest in megacities and where IP protection and policy execution are robust.
medium quasi_experimental openalex
Study on the Impact of Establishing Big Data Comprehensive Pilot Zones on Urban Economic Resilience · Peiyi He · Apr 08, 2026
Workflow-integrated foundation models turn meetings and documents into durable, queryable 'artifact capital', making AI-enabled remote-first arrangements the default for much knowledge work. Offices instead become a resource for apprenticeship, conflict repair and other high-tacitness activities, while new risks—surveillance, skill atrophy and carbon costs—demand targeted ML and policy research.
n/a commentary openalex
Remote-Capable Knowledge Work Should Default to AI-Enabled Flexibility · Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao · Apr 08, 2026
Start-ups that adopt big-data analytics face higher early failure rates because of greater costs and sales uncertainty, yet surviving adopters grow faster, earn more and are more likely to attract VC — suggesting big-data acts as a performance amplifier for firms that can manage the risk.
medium correlational openalex
Big data-based management decisions and start-up performance · Elisa Rodepeter, Christoph Gschnaidtner, Hanna Hottenrott · Apr 08, 2026
A single theory argues that high-performing distributed systems become maximally heterogeneous within environmental limits, with communication networks setting how that diversity is organised; applying this 'Principle of Maximum Heterogeneity' yields concrete redesign ideas for large-scale AI compute and other production systems.
n/a theoretical openalex
The Principle of Maximum Heterogeneity Optimises Productivity in Distributed Production Systems Across Biology, Economics, and Computing · Guillhem Artis, Danyal Akarca, Jascha Achterberg · Apr 08, 2026
A new pipeline turns real software into 10,000+ long-horizon agent environments tied to U.S. occupations, creating CUA-World for evaluating computer-use assistants; distilled 2B vision-language models and reviewer-audit loops yield measurable but modest performance gains on these economically relevant tasks.
n/a descriptive arxiv
Gym-Anything: Turn any Software into an Agent Environment · Pranjal Aggarwal, Graham Neubig, Sean Welleck · Apr 07, 2026
A modular agentic AI system, Flowr, cut manual coordination and tightened demand-supply alignment in a supermarket-chain pilot by automating decision-intensive supply-chain tasks while keeping managers in the loop; findings are promising but rest on a single-firm deployment without rigorous causal controls.
low descriptive arxiv
Flowr -- Scaling Up Retail Supply Chain Operations Through Agentic AI in Large Scale Supermarket Chains · Eranga Bandara, Ross Gore, Sachin Shetty, Piumi Siyambalapitiya, Sachini Rajapak... · Apr 07, 2026
Code-writing AI agents often pass tests but fail to respect project design standards: a benchmark of 495 real issues finds under half of auto-resolved patches meet repository-specific design constraints, and test success is a weak predictor of design compliance.
medium descriptive arxiv
Does Pass Rate Tell the Whole Story? Evaluating Design Constraint Compliance in LLM-based Issue Resolution · Kai Yu, Zhenhao Zhou, Junhao Zeng, Ying Wang, Xueying Du, Zhiqiang Yuan, Junwei ... · Apr 07, 2026
A simulated multi-agent marketplace shows that only a few LLM retailers reliably grow capital while most break even, revealing a winner-take-most pattern in economic task performance; semantic matching alone does not predict economic success.
medium descriptive arxiv
Market-Bench: Benchmarking Large Language Models on Economic and Trade Competition · Yushuo Zheng, Huiyu Duan, Zicheng Zhang, Yucheng Zhu, Xiongkuo Min, Guangtao Zha... · Apr 07, 2026
Countries with stronger AI-assisted peer review systems produce substantially more science: a one-standard-deviation rise in a new AI Review Capability Index corresponds to roughly an 18–25% jump in national scientific output, driven by faster, more reproducible evaluation processes.
medium correlational arxiv
AI-Augmented Peer Review and Scientific Productivity: A Cross-Country Panel and SEM Analysis · Dongsoo Han · Apr 07, 2026