Knexus Research (United States)

While alignment in artificial intelligence (AI) is broadly concerned with how AI systems align with human values, decision-maker alignment refers to how algorithms align with the values of individual decision makers. Consequently, the values targeted in decision-maker alignment are attributes that influence the decision-making process employed by humans. Example of a cognitive attribute is risk tolerance. The ideal environments for investigating decision-maker alignment are those in which the optimal decision may not be available, forcing humans to compromise and select a suboptimal decision. An optimal decision may not be available due to environment constraints such as limited resources, uncertainty, or some source of pressure. In contrast with supervised machine learning, decision-maker alignment can be investigated with labeled data in which decisions are made by different decision makers who are influenced by cognitive attributes. The problem is that datasets to study decision-maker alignment are difficult and expensive to create and they result in a small number of samples. For this reason, this paper proposes an approach to extend existing datasets for the purpose of studying decision-maker alignment. We exemplify our proposed approach by extending a dataset of health insurance alternatives.

This discussion examines segmentation by environmental attitudes in Kuwait using a survey (n = 1121) of adult Kuwaiti citizens. Attitudes were measured with a shortened version of the Environmental Attitudes Inventory (EAI) scale as adapted in prior work about the UAE. Consistent with research suggesting that environmental attitudes frequently reflect local conditions, the dimensional structure in Kuwait was not very similar to that of the UAE case. The Kuwait data do, however, yield a dimensional structure which reflects the broader debate in the sustainability literature on the relative balance of environmental vs. socio-economic concerns. Cluster analysis on these dimensions yielded five distinct segments ranging from not very strong concern about either to strong concern about both, for focusing on one or the other. Policy implications of addressing different views about environmentalism are discussed.

Abstract This study seeks to improve the quantification of greenhouse gas (GHG) emissions from unconventional natural gas production in the Appalachian Basin, covering Carbon Dioxide (CO2), Methane (CH4), and Nitrous Oxide (N2O). The goal is to automate the calculation process for compliance with the US Environmental Protection Agency (EPA), corporate sustainability reporting, and internal monitoring for emissions reduction targets. Traditionally, GHG emissions are manually computed once a year and are often calculated by third-party consultants, reducing transparency in calculation approaches. The development of an on-demand GHG emissions inventory enables real-time identification of emission-intensive processes, guiding strategic decisions for emissions mitigation and methane intensity reductions. In this project, we employed Python and SQL scripts to map activity data related to process and fugitive emissions, including pneumatic devices, drilling and completions combustion, equipment leaks, liquids unloading, dehydrators, and flaring, into US EPA Subpart W formulas. The result is an annualized carbon footprint calculated at the well pad level, providing comprehensive insights into component gasses (CO2, CH4, and N2O), production-based carbon and methane intensities, and emissions contributions by emission source. Automation improved the efficiency and accuracy of the emissions calculation workflow, generating an updated carbon footprint in a runtime under 5 minutes. Additionally, the inclusion of emissions-generating events overlooked due to human error substantially improved the quality of the emissions footprint. Lastly, the employment of data analytics on all calculation inputs and outputs aided in identifying outliers that required modification. The GHG calculation tool enhances accuracy, transparency, and consistency while minimizing the time invested in GHG estimations, allowing teams to redirect their efforts toward value-added activities. Next steps in the project include detailed emissions forecasting, which facilitates determining the most cost-effective emissions reduction strategy. The mapping of activity data into GHG estimation models was completed within six weeks, with additional time spent in result validation against previous manual calculations. After an initial configuration, data mappings do not require ongoing updates, establishing a fully automated GHG calculation process. This advancement enables efficiency and reliability in monitoring and reporting GHG emissions. Additionally, the development of a granular carbon footprint on an emission source basis enables detailed, source-level emissions forecasting to support future decisions around emission reduction projects.

We present an approach to algorithmic decision-making that emulates key facets of human decision-making, particularly in scenarios marked by expert disagreement and ambiguity. Our system employs a case-based reasoning framework, integrating learned experiences, contextual factors, probabilistic reasoning, domain-specific knowledge, and the personal traits of decision-makers. A primary aim of the system is to articulate algorithmic decision-making as a human-comprehensible reasoning process, complete with justifications for selected actions.

TU Tau (=HD 38218 = HIP 27135) is a binary system consisting of a C-N carbon star primary and an A-type secondary. We report on new photometry and spectroscopy which tracked the recent disappearance of the A-star secondary. The dimming of the A star was gradual and irregular, with one or more brief brightenings, implying the presence of nonhomogeneities in the carbon star outflow. We also present evidence that the A star is actively accreting s-process enriched material from the carbon star and suggest that it will therefore eventually evolve into a Barium giant. This is an important system as well because the A-type star can serve as a probe of the outer atmosphere of the carbon star.

The Census TopDown Algorithm (TDA) is a disclosure avoidance system using differential privacy for privacy-loss accounting. The algorithm ingests the final, edited version of the 2020 Census data and the final tabulation geographic definitions. The algorithm then creates noisy versions of key queries on the data, referred to as measurements, using zero-Concentrated Differential Privacy. Another key aspect of the TDA are invariants, statistics that the Census Bureau has determined, as matter of policy, to exclude from the privacy-loss accounting. The TDA postprocesses the measurements together with the invariants to produce a Microdata Detail File (MDF) that contains one record for each person and one record for each housing unit enumerated in the 2020 Census. The MDF is passed to the 2020 Census tabulation system to produce the 2020 Census Redistricting Data (P.L. 94-171) Summary File. This article describes the mathematics and testing of the TDA for this purpose.

Intelligent systems such as technologies related to artificial intelligence, robotics, machine learning, etc. open new insights into data and expand the concept of work in myriad domains. These technologies, while potentially useful, face high barriers to widespread adoption and acceptance by industries and citizens alike. The complexity and multi-dimensionality inherent in intelligent systems often renders traditional validation efforts (e.g., traceability analysis) impossible. In addition, contexts where predictions or computer-generated recommendations have real-world consequences, such as in medical prognosis, financial investing, or military applications introduce new risks and a host of moral and ethical concerns that can further hinder the widespread adoption of intelligent systems. Naturally, such reluctance by would-be users limits the potential of intelligent systems to solve real-world problems. This article reviews the challenges to technology acceptance through the lens of system transparency and user trust, and extends the Technology Acceptance Model (TAM) structure with issues germane to intelligent systems. We examine several prospective transparency frameworks that could be adopted and used by Human-Computer Interaction (HCI) practitioners involved in systems development. Our intention is to assist practitioners in the design of more transparent systems with a specific eye towards enhancing trust and acceptance in intelligent systems. Further, as a result of our review, we suggest that the well-known TAM should be expanded in the context of intelligent systems to include trust and transparency as key elements of the model. Finally, we conclude with a research agenda that might offer empirical evidence showing how transparency might enhance acceptance and use of intelligent systems.

PurposeMost of the research studies on entry mode have been done through the lens of manufacturing firms from mature economic countries entering the Global South countries, and most of the related frameworks and theories have been developed to explain this perspective. This paper presents an understanding of entry mode strategies of services firms from the Global South countries entering into countries with mature economies by considering six software services firms in India entering the US market. The study develops a framework that incorporates multiple theories – both from an emerging economy perspective as well as a services perspective.Design/methodology/approachThis study is exploratory in nature and focuses on the how and why aspects of strategic decisions. Qualitative research has traditionally been chosen when the main research objective is to improve understanding of a phenomenon. A multiple case study approach was used in this research. One of the key aspects of this study is the impact of different theories (institutional, network and other theories) on enterprises of different sizes. Therefore six cases – two each on small, medium and large enterprises – were selected.FindingsResults identify addressable market size, cultural aspects, firm size, resource and service characteristics as dominating factors that influenced choice of entry modes. From a theoretical perspective, the author finds that theories such as transaction cost theory, eclectic paradigm and other popular theories associated with entry mode do not successfully explain the entry mode strategies for firms from the Global South. Strategic theory, such as resource-based view, motivation of a firm, have some application but do not explain, in isolation, all aspects of the entry mode choices.Research limitations/implicationsThe study is an exploratory study and needs more data to validate the themes expressed in its conclusions. The study also focused on one industry and one country. Different industries within the same country may have different characteristics and may not follow the observations found in this study. Similarly, firms from other countries with similar economic characteristics may exhibit very different behavior. As a result, the themes that were expressed for software companies in India may not be generalizable across other industries and countries.Originality/valueThis study is significant for both academics and managers. For academics, the study addresses a significant gap in the literature. For managers, this study provides a framework for managers to evaluate and select the entry mode strategy that would be most effective in a successful expansion into foreign countries.