Despite advances in AI's performance and interpretability, AI advisors can undermine experts' decisions and increase the time and effort experts must invest to make decisions. Consequently, AI systems deployed in high-stakes settings often fail to consistently add value across contexts and can even diminish the value that experts alone provide. Beyond harm in specific domains, such outcomes impede progress in research and practice, underscoring the need to understand when and why different AI advisors add or diminish value. To bridge this gap, we stress the importance of assessing the value AI advice brings to real-world contexts when designing and evaluating AI advisors. Building on this perspective, we characterize key pillars -- pathways through which AI advice impacts value -- and develop a framework that incorporates these pillars to create reliable, personalized, and value-adding advisors. Our results highlight the need for system-level, value-driven development of AI advisors that advise selectively, are tailored to experts' unique behaviors, and are optimized for context-specific trade-offs between decision improvements and advising costs. They also reveal how the lack of inclusion of these pillars in the design of AI advising systems may be contributing to the failures observed in practical applications.

Many machine learning models are susceptible to adversarial attacks, with decision-based black-box attacks representing the most critical threat in real-world applications. These attacks are extremely stealthy, generating adversarial examples using hard labels obtained from the target machine learning model. This is typically realized by optimizing perturbation directions, guided by decision boundaries identified through query-intensive exact search, significantly limiting the attack success rate. This paper introduces a novel approach using the Approximation Decision Boundary (ADB) to efficiently and accurately compare perturbation directions without precisely determining decision boundaries. The effectiveness of our ADB approach (ADBA) hinges on promptly identifying suitable ADB, ensuring reliable differentiation of all perturbation directions. For this purpose, we analyze the probability distribution of decision boundaries, confirming that using the distribution's median value as ADB can effectively distinguish different perturbation directions, giving rise to the development of the ADBA-md algorithm. ADBA-md only requires four queries on average to differentiate any pair of perturbation directions, which is highly query-efficient. Extensive experiments on six well-known image classifiers clearly demonstrate the superiority of ADBA and ADBA-md over multiple state-of-the-art black-box attacks. The source code is available at https://github.com/BUPTAIOC/ADBA.

Expert decision-makers (DMs) in high-stakes AI-assisted decision-making (AIaDM) settings receive and reconcile recommendations from AI systems before making their final decisions. We identify distinct properties of these settings which are key to developing AIaDM models that effectively benefit team performance. First, DMs incur reconciliation costs from exerting decision-making resources (e.g., time and effort) when reconciling AI recommendations that contradict their own judgment. Second, DMs in AIaDM settings exhibit algorithm discretion behavior (ADB), i.e., an idiosyncratic tendency to imperfectly accept or reject algorithmic recommendations for any given decision task. The human's reconciliation costs and imperfect discretion behavior introduce the need to develop AI systems which (1) provide recommendations selectively, (2) leverage the human partner's ADB to maximize the team's decision accuracy while regularizing for reconciliation costs, and (3) are inherently interpretable. We refer to the task of developing AI to advise humans in AIaDM settings as learning to advise and we address this task by first introducing the AI-assisted Team (AIaT)-Learning Framework. We instantiate our framework to develop TeamRules (TR): an algorithm that produces rule-based models and recommendations for AIaDM settings. TR is optimized to selectively advise a human and to trade-off reconciliation costs and team accuracy for a given environment by leveraging the human partner's ADB. Evaluations on synthetic and real-world benchmark datasets with a variety of simulated human accuracy and discretion behaviors show that TR robustly improves the team's objective across settings over interpretable, rule-based alternatives.

The Asian Development Bank (ADB) was formed with the aim of fostering economic growth in some of the poorest regions in the world. With that as its charter, ADB's CIO from the organisation's office of information systems and technology, Shirin Hamid, said it's difficult to compare ADB to the likes of other banks when talking about digital transformation. "DBS for example in Singapore has basically positioned itself as a digital bank, but as a development bank, where our partners and what we deliver is on the ground with countries that are from different landscapes -- we have middle income and the least developing countries -- are they ready for a bank like us to be a digital bank?," she said, speaking at Gartner IT Symposium/Xpo on the Gold Coast last week. You can download all of the articles in this special report in one PDF (free registration required). Hamid said she recently had a conversation with ADB's vice president of finance and risk management centred on determining if ADB was ready to be a digital bank.

The belief that robots, automation, and AI simply displace jobs and make humans irrelevant is not borne out in Asia, reports Chris Middleton. However, there are lessons to learn from the technologies' impact – in Asia and the rest of the world. Robots and automation are creating more jobs in Asia than they destroy, according to a new report from the Asia Development Bank (ADB). ADB analysis of a dozen Asian economies between 2005 and 2015 found that rising demand more than compensated for jobs lost to automation. The adoption of robotics and other connected systems stimulated higher productivity and economic growth, creating 134 million new jobs, compared with the 101 million lost to new technologies.

NEW DELHI: As newer technologies like robotics and artificial intelligence (AI) take over, workers need to upskill or face the risk of lower wages and unemployment, ADB warned today. ADB's'Asian Development Outlook, 2018' has called for government action on skill development coupled with steps in labour regulation, social protection and income redistribution. It said jobs which require repetitive, routine tasks and workers who do not have the education or training to move easily to other occupation, may face slow growth in wages. "New technologies alter the skills required of the workforce and may cause unemployment as some firms downsize or close. They make the less-skilled more likely to experience lower wage growth, exacerbating income inequality," it said.

A Bengaluru-based company has come up with a solution that could help the Indian Army leap over the language barrier that has been impeding communication between soldiers deployed in forward areas and their Chinese counterparts. Cogknit Semantics Pvt Ltd has offered the army an artificial intelligence (AI)-powered solution that can translate Chinese into English or Hindi in real time, according to a new report prepared by the Federation of Indian Chambers of Commerce and Industry (FICCI) on solutions to problems identified by the army. Army chief General Bipin Rawat released the report, titled '1st Compendium on Solutions to Problem Statements,' at the DefExpo-2018, which concluded in Chennai on Saturday. The Chinese language barrier is one of the 130 problems identified by the Army Design Bureau (ADB) in three separate reports. Other problems include a drop in engine performance of tanks and infantry combat vehicles at high altitude, and the difficulties in laying bridges for movement of troops and vehicles in mountains.