Search for: All records

Nudging is a behavioral strategy aimed at influencing people’s thoughts and actions. Nudging techniques can be found in many situations in our daily lives, and these nudging techniques can targeted at human fast and unconscious thinking, e.g., by using images to generate fear or the more careful and effortful slow thinking, e.g., by releasing information that makes us reflect on our choices. In this paper, we propose and discuss a value-based AI-human collaborative framework where AI systems nudge humans by proposing decision recommendations. Three different nudging modalities, based on when recommendations are presented to the human, are intended to stimulate human fast thinking, slow thinking, or meta-cognition. Values that are relevant to a specific decision scenario are used to decide when and how to use each of these nudging modalities. Examples of values are decision quality, speed, human upskilling and learning, human agency, and privacy. Several values can be present at the same time, and their priorities can vary over time. The framework treats values as parameters to be instantiated in a specific decision environment. 

Voting is used widely to identify a collective decision for a group of agents, based on their preferences. In this paper, we focus on evaluating and designing voting rules that support both the privacy of the voting agents and a notion of fairness over such agents. To do this, we introduce a novel notion of group fairness and adopt the existing notion of local differential privacy. We then evaluate the level of group fairness in several existing voting rules, as well as the trade-offs between fairness and privacy, showing that it is not possible to always obtain maximal economic efficiency with high fairness or high privacy levels. Then, we present both a machine learning and a constrained optimization approach to design new voting rules that are fair while maintaining a high level of economic efficiency. Finally, we empirically examine the effect of adding noise to create local differentially private voting rules and discuss the three-way trade-off between economic efficiency, fairness, and privacy.This paper appears in the special track on AI & Society. 

Many real-life scenarios require humans to make difficult trade-offs: do we always follow all the traffic rules or do we violate the speed limit in an emergency? In general, how should we account for and balance the ethical values, safety recommendations, and societal norms, when we are trying to achieve a certain objective? To enable effective AI-human collaboration, we must equip AI agents with a model of how humans make such trade-offs in environments where there is not only a goal to be reached, but there are also ethical constraints to be considered and to possibly align with. These ethical constraints could be both deontological rules on actions that should not be performed, or also consequentialist policies that recommend avoiding reaching certain states of the world. Our purpose is to build AI agents that can mimic human behavior in these ethically constrained decision environments, with a long term research goal to use AI to help humans in making better moral judgments and actions. To this end, we propose a computational approach where competing objectives and ethical constraints are orchestrated through a method that leverages a cognitive model of human decision making, called multi-alternative decision field theory (MDFT). Using MDFT, we build an orchestrator, called MDFT-Orchestrator (MDFT-O), that is both general and flexible. We also show experimentally that MDFT-O both generates better decisions than using a heuristic that takes a weighted average of competing policies (WA-O), but also performs better in terms of mimicking human decisions as collected through Amazon Mechanical Turk (AMT). Our methodology is therefore able to faithfully model human decision in ethically constrained decision environments. 

Currents are unique drivers of oceanic phylogeography and thus determine the distribution of marine coastal species, along with past glaciations and sea-level changes. Here we reconstruct the worldwide colonization history of eelgrass (Zostera marinaL.), the most widely distributed marine flowering plant or seagrass from its origin in the Northwest Pacific, based on nuclear and chloroplast genomes. We identified two divergent Pacific clades with evidence for admixture along the East Pacific coast. Two west-to-east (trans-Pacific) colonization events support the key role of the North Pacific Current. Time-calibrated nuclear and chloroplast phylogenies yielded concordant estimates of the arrival ofZ. marinain the Atlantic through the Canadian Arctic, suggesting that eelgrass-based ecosystems, hotspots of biodiversity and carbon sequestration, have only been present there for ~243 ky (thousand years). Mediterranean populations were founded ~44 kya, while extant distributions along western and eastern Atlantic shores were founded at the end of the Last Glacial Maximum (~19 kya), with at least one major refuge being the North Carolina region. The recent colonization and five- to sevenfold lower genomic diversity of the Atlantic compared to the Pacific populations raises concern and opportunity about how Atlantic eelgrass might respond to rapidly warming coastal oceans.

 

Current AI systems lack several important human capabilities, such as adaptability, generalizability, selfcontrol, consistency, common sense, and causal reasoning. We believe that existing cognitive theories of human decision making, such as the thinking fast and slow theory, can provide insights on how to advance AI systems towards some of these capabilities. In this paper, we propose a general architecture that is based on fast/slow solvers and a metacognitive component. We then present experimental results on the behavior of an instance of this architecture, for AI systems that make decisions about navigating in a constrained environment. We show how combining the fast and slow decision modalities, which can be implemented by learning and reasoning components respectively, allows the system to evolve over time and gradually pass from slow to fast thinking with enough experience, and that this greatly helps in decision quality, resource consumption, and efficiency. 

A cobalt oxide (Co3O4)-decorated silicon carbide (SiC) nano-tree array (denoted as Co3O4/SiC NTA) electrode is synthesized, and it is investigated for use in micro-supercapacitor applications. Firstly, the well-standing SiC nanowires (NWs) are prepared by nickel (Ni)-catalyzed chemical vapor deposition (CVD) method, and then the thin layer of Co3O4 and the hierarchical Co3O4 nano-flower-clusters are, respectively, fabricated on the side-walls and the top side of the SiC NWs via electrodeposition. The deposition of Co3O4 on the SiC NWs benefits the charge transfer at the electrode/aqueous electrolyte interface due to its extremely hydrophilic surface characteristic after Co3O4 decoration. Furthermore, the Co3O4/SiC NTA electrode provides a directional charge transport route along the length of SiC nanowires owing to their well-standing architecture. By using the Co3O4/SiC NTA electrode for micro-supercapacitor application, the areal capacitance obtained from cyclic voltammetry measurement reaches 845 mF cm−2 at a 10 mV s−1 scan rate. Finally, the capacitance durability is also evaluated by the cycling test of cyclic voltammetry at a high scan rate of 150 mV s−1 for 2000 cycles, exhibiting excellent stability. 

Many real-life scenarios require humans to make difficult trade-offs: do we always follow all the traffic rules or do we violate the speed limit in an emergency? These scenarios force us to evaluate the trade-off between collective norms and our own personal objectives. To create effective AI-human teams, we must equip AI agents with a model of how humans make trade-offs in complex, constrained environments. These agents will be able to mirror human behavior or to draw human attention to situations where decision making could be improved. To this end, we propose a novel inverse reinforcement learning (IRL) method for learning implicit hard and soft constraints from demonstrations, enabling agents to quickly adapt to new settings. In addition, learning soft constraints over states, actions, and state features allows agents to transfer this knowledge to new domains that share similar aspects. 

Distribution of Earth’s biomes is structured by the match between climate and plant traits, which in turn shape associated communities and ecosystem processes and services. However, that climate–trait match can be disrupted by historical events, with lasting ecosystem impacts. As Earth’s environment changes faster than at any time in human history, critical questions are whether and how organismal traits and ecosystems can adjust to altered conditions. We quantified the relative importance of current environmental forcing versus evolutionary history in shaping the growth form (stature and biomass) and associated community of eelgrass ( Zostera marina ), a widespread foundation plant of marine ecosystems along Northern Hemisphere coastlines, which experienced major shifts in distribution and genetic composition during the Pleistocene. We found that eelgrass stature and biomass retain a legacy of the Pleistocene colonization of the Atlantic from the ancestral Pacific range and of more recent within-basin bottlenecks and genetic differentiation. This evolutionary legacy in turn influences the biomass of associated algae and invertebrates that fuel coastal food webs, with effects comparable to or stronger than effects of current environmental forcing. Such historical lags in phenotypic acclimatization may constrain ecosystem adjustments to rapid anthropogenic climate change, thus altering predictions about the future functioning of ecosystems. 

While considerable evidence exists of biogeographic patterns in the intensity of species interactions, the influence of these patterns on variation in community structure is less clear. Studying how the distributions of traits in communities vary along global gradients can inform how variation in interactions and other factors contribute to the process of community assembly. Using a model selection approach on measures of trait dispersion in crustaceans associated with eelgrass ( Zostera marina ) spanning 30° of latitude in two oceans, we found that dispersion strongly increased with increasing predation and decreasing latitude. Ocean and epiphyte load appeared as secondary predictors; Pacific communities were more overdispersed while Atlantic communities were more clustered, and increasing epiphytes were associated with increased clustering. By examining how species interactions and environmental filters influence community structure across biogeographic regions, we demonstrate how both latitudinal variation in species interactions and historical contingency shape these responses. Community trait distributions have implications for ecosystem stability and functioning, and integrating large-scale observations of environmental filters, species interactions and traits can help us predict how communities may respond to environmental change.