Editor's note: This is the third in a series of articles about the effects of software and automation on the economy. You can read the other stories here and here. The way Hod Lipson describes his Creative Machines Lab captures his ambitions: "We are interested in robots that create and are creative." Lipson, an engineering professor at Cornell University (this July he's moving his lab to Columbia University), is one of the world's leading experts on artificial intelligence and robotics. His research projects provide a peek into the intriguing possibilities of machines and automation, from robots that "evolve" to ones that assemble themselves out of basic building blocks. A few years ago, Lipson demonstrated an algorithm that explained experimental data by formulating new scientific laws, which were consistent with ones known to be true. He had automated scientific discovery. Lipson's vision of the future is one in which machines and software possess abilities that were unthinkable until recently.

John Hennessy joined Stanford in 1977 right after receiving his Ph.D. from the State University of New York at Stony Brook. He soon became a leader of Reduced Instruction Set Computers. This research led to the founding of MIPS Computer Systems, which was later acquired for 320 million. There are still nearly a billion MIPS processors shipped annually, 30 years after the company was founded. Hennessy returned to Stanford to do foundational research in large-scale shared memory multiprocessors. In his spare time, he co-authored two textbooks on computer architecture, which have been continuously revised and are still popular 25 years later. This record led to numerous honors, including ACM Fellow, election to both the National Academy of Engineering and the National Academy of Sciences. Not resting on his research and teaching laurels, he quickly moved up the academic administrative ladder, going from the CS department chair to Engineering college dean to provost and finally to president in just seven years. He is Stanford's tenth president, its first from engineering, and he has governed it for an eighth of its existence. Since 2000, he doubled Stanford's endowment, including a record 6.2 billion for a single campaign. He used those funds to launch many initiatives--which often cross departmental lines--along with new buildings to house them. Undergraduate applications also doubled, for the first time making Stanford even more selective than Harvard.

It described as "fundamentally vague" Stephen Goose's ethical line in his Point side of the Point/Counterpoint debate "The Case for Banning Killer Robots" in the same issue. I encourage all ACM members to read or re-read them and consider if they themselves should be working on lethal autonomous weapons or even on any kind of weapon. Ronald Arkin's Counterpoint was optimistic regarding robots' ability to "... exceed human moral performance ...," writing that a ban on autonomous weapons "... ignores the moral imperative to use technology to reduce the atrocities and mistakes that human warfighters make." This analysis involved two main problems. First, Arkin tacitly assumed autonomous weapons will be used only by benevolent forces, and the "moral performance" of such weapons is incorruptible by those deploying them.

We study large-scale kernel methods for acoustic modeling and compare to DNNs on performance metrics related to both acoustic modeling and recognition. Measuring perplexity and frame-level classification accuracy, kernel-based acoustic models are as effective as their DNN counterparts. However, on token-error-rates DNN models can be significantly better. We have discovered that this might be attributed to DNN's unique strength in reducing both the perplexity and the entropy of the predicted posterior probabilities. Motivated by our findings, we propose a new technique, entropy regularized perplexity, for model selection. This technique can noticeably improve the recognition performance of both types of models, and reduces the gap between them. While effective on Broadcast News, this technique could be also applicable to other tasks.

Traditional theories of social models conceptualize teams as distributed processors, disregarding the interdependence necessary to multi-task. Yet, interdependence characterizes social behavior. Instead, traditional theory favor cooperation, a state of least entropy production (LEP), without understanding the causes, limits or consequences of cooperation. As a simple example of interdependence, foraging prey overgraze forests free of predators. In our model, interdependence creates uncertainty, tradeoffs and signals (e.g., prices, coordination, innovation). Unlike individuals, the ability of teams to multitask reflects a quantum-like entanglement that represents maximum entropy production (MEP) when solving the problems signaled by society to improve its welfare. Our model supports findings that evolution in nature is driven by the MEP from making intelligent choices. Exploiting interdependence improves team intelligence, improves performance and reduces the risk of human error; forced cooperation disorganizes it by increasing the risk of error; e.g., if team cooperation improves teamwork, widespread forced cooperation in an autocracy or bureaucracy reduces social intelligence by adding unnecessary noise to signals. In our model, competition between teams self-organizes outsiders willing to sort through the noise for signals of the choices that improve social welfare (e.g., teams in courtrooms; science; entertainment; sports; businesses). Social systems organized around competition (e.g., stronger signals from robust checks and balances) better control a society by more correctly sizing teams to solve problems with fewer errors compared to autocracies or bureaucracies. Overall, we predict, the density of MEP directed at solving problems in a society with the constraints imposed from strong checks and balances, yet able to freely self-organize its labor and capital within those constraints, is denser.

We study the complexity of a combinatorial variant of the Shift Bribery problem in elections. In the standard Shift Bribery problem, we are given an election where each voter has a preference order over the set of candidates and where an outside agent, the briber, can pay each voter to rank the briber's favorite candidate a given number of positions higher. The goal is to ensure the victory of the briber's preferred candidate. The combinatorial variant of the problem, introduced in this paper, models settings where it is possible to affect the position of the preferred candidate in multiple votes, either positively or negatively, with a single bribery action. This variant of the problem is particularly interesting in the context of large-scale campaign management problems (which, from the technical side, are modeled as bribery problems). We show that, in general, the combinatorial variant of the problem is highly intractable; specifically, NP-hard, hard in the parameterized sense, and hard to approximate. Nevertheless, we provide parameterized algorithms and approximation algorithms for natural restricted cases.

This paper and use of autonomous weapons systems has been will discuss the moral and ethical questions that arise in the one of the outcomes of the counterterrorism and counterinsurgency use of lethally autonomous technology for military purposes operations in Iraq and Afghanistan. The asymmetrical and how the forms of subjectivity and moral agency that battlefields of these theaters, where no frontline it creates could be highly counterproductive to mission provides a buffer between combatants and civilians and effectiveness, diplomacy and conflict resolution and prevention.

This paper proposes a model for an artificial autonomous moral agent (AAMA), which is parsimonious in its ontology and minimal in its ethical assumptions. Starting from a set of moral data, this AAMA is able to learn and develop a form of moral competency. It resembles an “optimizing predictive mind,” which uses moral data (describing typical behavior of humans) and a set of dispositional traits to learn how to classify different actions (given a given background knowledge) as morally right, wrong, or neutral. When confronted with a new situation, this AAMA is supposedly able to predict a behavior consistent with the training set. This paper argues that a promising computational tool that fits our model is “neuroevolution,” i.e. evolving artificial neural networks.

It is nowadays our responsibility to convince our contemporary citizens that AI devices as UGVs (Unmanned Ground Vehicles) and UAVs (Unmanned Aerial Vehicles) are crucial actors of today’s life in a dual domains, both civilian and military. In particular, the decision process is the main component of every military operation and is of high interest because of two main reasons : it is necessary designed to cope with conflict issues and it requires a very complex planning process to be successful. The difficulty to find a good plan is worse in urban areas because of the high uncertainty due to the topology of these areas, the presence of civilians, who can be hostile or friendly, and the unpredictable nature of enemies. The idea in that paper is to qualify what can be a valid computed plan in that context , i.e. welldesigned for recovering of peace, rescue operations after a bombing event, hostage salvage, non-combatant evacuation operations, civil-military co-operation, ...., in urban areas. This planning process leads to associate actually four components, the representation of the tactical scheme, the implementation of the tactical scheme as the behaviour of special forces, military units or emergency squads, the proof process or the explanation process, and finally the handling of external factors depending on the current environment or the current context in which the operation takes place. This paper uses a quaternary representation called the epistemological quadriptych, in order to highlight that the integration of UGVs or UAVs devices requires actually to understand the role of knowledge and behaviour and to provide secure and valid action plans, i.e. which can be explained and justified.

However, foundational problems remain in the either mindfully or inadvertently by individuals or teams of continuing development of AI for team autonomy, humans. One worry about this bright future is that jobs especially with objective measures able to optimize team may be lost; from Mims (2015), function, performance and composition. Something potentially momentous is happening inside AI approaches often attempt to address autonomy by startups, and it's a practice that many of their established modeling aspects of human decision-making or behavior.