This paper reveals how location-based recommendation apps, like those suggesting restaurants, can unintentionally increase income segregation in cities and cause many local businesses to disappear. As these apps get "smarter," they often push people into more homogenous social circles, a long-term effect missed by how we usually measure their success.
Recommendation Feedback Loops Shape Urban Income Segregation and Venue Visitation · Guangyuan Weng, Lin Chen, Ciro Beneduce et al. · arXiv:2610.41911
AgenticMobility is an open-source platform that simulates AI agents acting as travellers and city managers, learning over time to shape urban traffic. It allows testing of how policies or individual behaviors might play out in future smart city planning and traffic management.
AgenticMobility: An Open-Source Platform for Multi-AI-Agent Mobility Dynamics · Jiaheng Xiao, Jiangtao Han, Jiachen Wang et al. · arXiv:2611.01918
The method fills in missing location data by using the fading influence of a person's past and future movements. Complete mobility data is essential for tracking disease spread and designing urban spaces.
Inferring missing locations in human mobility data using memory effects · Pablo Rosillo-Rodes, Antònia Tugores, David Sánchez et al. · arXiv:2611.01319
SAGE uses large language models to discover rules for how complex networks such as brain activity or social groups change over time. It can tell general group behavior from individual variation and propose new rules when needed.
Discovering hierarchical governing equations for community-structured complex network dynamics with LLM agents · Jinlei Zhang, Zixuan Zhang, Hao Qiu et al. · arXiv:2611.02816
Promised small annual power cuts can hide a need for large facilities to fully disconnect for short periods. Data center growth is outpacing new power generation, so realistic expectations are needed for flexible demand.
Siting and Curtailment Depth Limit the Grid Headroom Available to New Large Loads · Nils Johnson, Geoff Blanford, John E. T. Bistline · arXiv:2611.00462
A model explores how humanoid robot numbers might grow compared with humans, including a future where robots help make more robots. It offers a way to think about the long-term societal effects of advanced AI and robotics, not to forecast.
When will androids outnumber humans? · Alessio Zaccone, Thomas Tunstall, Jesse H. Ausubel · arXiv:2611.03242
General-purpose AI models forecasted electricity carbon intensity in Nordic regions over four days more accurately than traditional methods. Better forecasts could help energy planners cut emissions and manage grids more efficiently.
Zero-shot time-series foundation models against the Khepri v1.5 baselines for 96-hour carbon-intensity forecasts in nine Nordic bidding zones: a preregistered comparison · Eirik Botten Nicolaysen · arXiv:2611.02895
A detailed test of confidence accuracy finds current AI forecasting models often systematically misjudge how sure they are. As people rely on these models for critical decisions, standard checks might mislead about their reliability.
Evaluating Uncertainty Calibration in Probabilistic Time Series Foundation Models · Thomas Decker, Volker Tresp · arXiv:2611.03592
Small random changes, such as a few more fish born or caught, can jiggle shared resources and amplify cycles. It helps explain persistent cycles in shared resources like global fish landings.
Resonant Amplification of Demographic Noise in Stochastic Common-Pool Resource Dynamics · Susanna C. Caruso, Fabio Menegazzo, Vito Latora et al. · arXiv:2610.41830
A look at how the way people are grouped in different parts of their lives, like at work versus with family, influences how diseases spread. If these groups are very separate, outbreaks can be intense but quick, while more overlapping groups lead to milder but longer epidemics.
Optimal Strategies to Contain and Prevent the Spread of Diseases in Group-structured Populations · Varun K. Rao, Siddharth Patwardhan, Filippo Radicchi · arXiv:2611.01341