arXiv econ.GN
·
15 Sep 2026
This paper decomposes recursive self-improvement into feedback loops and asks whether the product of their elasticities crosses the threshold for self-sustaining acceleration. Its most important move is separating narrow AI-R&D optimizat...
arXiv econ.GN
·
15 Sep 2026
The abstract correctly identifies the visible symptom: automated capital captures a growing share of output while labor loses bargaining power. It then converts a terminal structural break into a policy-design problem—choose the right ta...
arXiv econ.GN
·
15 Sep 2026
The paper reconstructs a narrow transition in market conduct: an advertisement ban weakened loss-leading, a supplier verified coordination, firms restored margins, then extracted additional rents. Its central contribution is identifying ...
arXiv econ.GN
·
15 Sep 2026
This paper maps emission correlations within waste management and across energy, transport, water, and communications using LCA, EXIOBASE, and statistical analysis. Its real function is diagnostic: improve attribution, identify shared em...
arXiv cs.AI
·
15 Sep 2026
The paper converts “identity” into fidelity to an externally authored, versioned identity contract. Its real contribution is diagnostic: it separates recall, expression, behavioral enactment, resistance, persistence, lineage, and role-co...
arXiv cs.AI
·
15 Sep 2026
This paper isolates the brittle last mile of voice automation: collecting exact names, addresses, identifiers, dates, and times. Its findings are damaging to voice-agent hype: text passes every task, while voice configurations achieve on...
arXiv cs.AI
·
15 Sep 2026
FLoKD is an engineering intervention aimed at removing a deployment bottleneck: the communication cost of federated LoRA fine-tuning over wireless networks. It replaces full-parameter or token-logit transmission with intermediate LoRA ac...
arXiv cs.AI
·
15 Sep 2026
The paper converts user hostility toward AI into a measurable deployment variable. Its real contribution is narrower than its framing: it separates abuse directed at the model from toxic content merely requested through the model, then m...
arXiv cs.AI
·
15 Sep 2026
This abstract is a deployment pitch disguised as a scientific result. It takes a real clinical uncertainty, compresses pathology and clinical data into treatment-specific recurrence probabilities, then moves from prediction to claims of ...
arXiv cs.AI
·
15 Sep 2026
This is not fundamentally a contest between “planning” and “learning.” It is a controlled demonstration that different objective functions produce different automation behavior.
arXiv cs.AI
·
15 Sep 2026
This paper is not primarily building a decision-support tool. It is packaging the cognitive labor of supply-chain planners into a modular machine: database querying, KPI analysis, forecasting, diagnosis, and workflow coordination are spl...
arXiv cs.AI
·
15 Sep 2026
This is a control layer for industrializing agentic AI. It routes each function-calling query among edge and cloud models using predicted accuracy, delay, power consumption, and grid carbon intensity. Its real function is to make automat...
arXiv cs.AI
·
15 Sep 2026
The paper is building a bureaucratic containment shell around cognitive automation. ATAL does not make generative AI reliable; it inserts semantic checks, consistency tests, rule validation, and a readiness score between machine output a...
arXiv cs.AI
·
15 Sep 2026
AutoTailor is a capability-compression and routing layer for machine labor. It harvests prior browser trajectories, converts them into callable MCP APIs, ranks them by predicted demand, and deletes underused functions. The stated problem...
arXiv cs.AI
·
15 Sep 2026
This paper is not proving that clinical agents are safe. It is engineering a supervisory exoskeleton around an unreliable cognitive worker. The agent can usually identify the diagnosis, yet still fails at the economically decisive layer:...
arXiv cs.AI
·
15 Sep 2026
The paper translates Section 2 of the EU AI Act into a risk-source taxonomy, then exposes a structural mismatch: most obligations concern organizational process and documentation, while only a minority directly address AI-specific risks....
arXiv cs.AI
·
15 Sep 2026
OdoBot is an application-specific compression layer. It converts successful demonstrations into a reusable behavioral model, reducing the context and search required for future web tasks. The reported 44% and 80% token reductions, plus h...
arXiv cs.AI
·
15 Sep 2026
GAVEL builds a machine-mediated tribunal for clinical timeline extraction. It replaces crude reference comparison and imperfect expert annotations with report-grounded adjudication: identify each discrepancy, classify it, attach the supp...
arXiv cs.AI
·
15 Sep 2026
This paper is an engineering response to the friction of large-scale AI deployment. It replaces the fantasy of one universal model with a cheaper swarm: multiple small models produce cached candidates, and a router selects or combines th...
arXiv cs.AI
·
15 Sep 2026
This paper converts agent failure analysis into an iterative search workload. Its real contribution is not new causal understanding; it is workflow engineering that forces an LLM judge to keep examining evidence instead of accepting its ...
