Racing to the precipice: a model of artificial intelligence development

This paper presents a simple model of an AI (artificial intelligence) arms race, where several development teams race to build the first AI. Under the assumption that the first AI will be very powerful and transformative, each team is incentivised to finish first—by skimping on safety precautions if need be. This paper presents the Nash equilibrium of this process, where each team takes the correct amount of safety precautions in the arms race. Having extra development teams and extra enmity between teams can increase the danger of an AI disaster, especially if risk-taking is more important than skill in developing the AI. Surprisingly, information also increases the risks: the more teams know about each others’ capabilities (and about their own), the more the danger increases. Should these results persist in more realistic models and analysis, it points the way to methods of increasing the chance of the safe development of AI.     

Work hours, workload, sleep and fatigue in Australian Rail Industry employees

Research suggests that less than 5 h sleep in the 24 h prior to work and/or more than 16 h of wakefulness can significantly increase the likelihood of fatigue-related impairment and error at work. Studies have also shown exponential safety declines with time on shift, with roughly double the likelihood of accident or injury after 10 h relative to the first 8 h. While it is acknowledged that reduced sleep, increased wakefulness and longer work hours produce work-related fatigue, few studies have examined the impact of workload on this relationship. Studies in the rail industry have focused on drivers. This study investigated fatigue in a large sample of Australian Rail Industry Employees. Participants were from four companies (n = 90: 85m, 5f; mean age 40.2 ± 8.6 y). Data was analysed for a total of 713 shifts. Subjects wore wrist actigraphs and completed sleep and work diaries for 14-days. They also completed the Samn–Perelli Fatigue Scale at the beginning and end of shifts, and the NASA-TLX workload scale at least twice during each shift. Average (±SD) sleep length (7.2 ± 2.6 h), prior wake at shift end (12.0 ± 4.7 h), shift duration (8.0 ± 1.3) and fatigue (4.1 ± 1.3, “a little tired, less than fresh”) were within limits generally considered acceptable from a fatigue perspective. However, participants received 5 h or less sleep in the prior 24 h on 13%, were awake for at least 16 h at the end of 16% and worked at least 10 h on 7% of shifts. Subjects reported that they felt “extremely tired, very difficult to concentrate,” or “completely exhausted, unable to function effectively” on 13% of shifts. Sleep length (OR = 0.88, p < 0.01), shift duration (OR = 1.18, p < 0.05), night shift (REF = morning shift, OR = 2.12, p < 0.05) and workload ratings (OR = 1.2, p < 0.05) were significant predictors of ratings of extreme tiredness/exhaustion (yes/no). While on average, sleep loss, extended wakefulness, longer work hours and work-related fatigue do not appear problematic in this sample, there is still a notable percentage of shifts that are likely to be associated with high levels of work-related fatigue. Given the size of the Australian Rail Industry, with thousands of shifts occurring each day, this is potentially of operational concern. Further, results indicate that, in addition to sleep length, wakefulness and work hours, workload significantly influences fatigue. This has possible implications for bio-mathematical predictions of fatigue and for fatigue management more generally.     

The minimum positional error incurred by any connectivity-based positioning algorithm for mobile wireless systems

A very important problem in many wireless ad-hoc networks, including wireless sensor networks, is positioning or the determination of geographical locations of the wireless nodes. Positioning is used both in infrastructural aspects of sensor networks, like geographic routing and topology maintenance, and in applications like wildlife tracking. Connectivity-based positioning algorithms in mobile wireless systems are studied in this work. These algorithms compute node positions based only on the connectivity, i.e. the neighborhood information of each node. Many algorithms have been proposed for positioning in stationary node systems and bounds on positional error of algorithms have been derived. The design and analysis of positioning algorithms for mobile node systems is a more challenging problem. Node mobility increases the amount of positional information available to a positioning algorithm. The work in this paper establishes a bound on the positional error for connectivity-based algorithms in mobile systems. The formulation from the analysis is used to investigate the benefit of this additional positional information on reducing positional error. There is a limit to the usefulness of positional information from previous node positions due to movement. This captures an important performance tradeoff: historical positional information can yield reduced positional error but requires more connectivity information from the network which requires greater computational resources.     

Moblin.org开源项目

市场上出现了一种新的设备,它能够访问Internet上丰富的内容,同时没有与PC相关的复杂性。这些设备以各种形式出现,从移动互联网设备（Mobile Internet Devices,MID）、便携式媒体播放器（PMP）到个人导航设备（PND）、数码机顶盒（STB）到笔记本电脑。Moblin.org项目由Intel于2007年7月正式启动,它的目的是推动此类设备所需的最新技术的发展。     

Mitochondria: A mirror into cellular dysfunction in heart disease

Cardiovascular (CV) disease is the single most significant cause of morbidity and mortality worldwide. The emerging global impact of CV disease means that the goals of early diagnosis and a wider range of treatment options are now increasingly pertinent. As such, there is a greater need to understand the molecular mechanisms involved and potential targets for intervention. Mitochondrial function is important for physiological maintenance of the cell, and when this function is altered, the cell can begin to suffer. Given the broad range and significant impacts of the cellular processes regulated by the mitochondria, it becomes important to understand the roles of the proteins associated with this organelle. Proteomic investigations of the mitochondria are hampered by the intrinsic properties of the organelle, including hydrophobic mitochondrial membranes; high proportion of basic proteins (pI greater than 8.0); and the relative dynamic range issues of the mitochondria. For these reasons, many proteomic studies investigate the mitochondria as a discrete subproteome. Once this has been achieved, the alterations that result in functional changes with CV disease can be observed. Those alterations that lead to changes in mitochondrial function, signaling and morphology, which have significant implications for the cardiomyocyte in the development of CV disease, are discussed.     

Practical considerations for high spatial and temporal resolution dynamic transmission electron microscopy

Although recent years have seen significant advances in the spatial resolution possible in the transmission electron microscope (TEM), the temporal resolution of most microscopes is limited to video rate at best. This lack of temporal resolution means that our understanding of dynamic processes in materials is extremely limited. High temporal resolution in the TEM can be achieved, however, by replacing the normal thermionic or field emission source with a photoemission source. In this case the temporal resolution is limited only by the ability to create a short pulse of photoexcited electrons in the source, and this can be as short as a few femtoseconds. The operation of the photo-emission source and the control of the subsequent pulse of electrons (containing as many as 5 x 10(7) electrons) create significant challenges for a standard microscope column that is designed to operate with a single electron in the column at any one time. In this paper, the generation and control of electron pulses in the TEM to obtain a temporal resolution <10(-6)s will be described and the effect of the pulse duration and current density on the spatial resolution of the instrument will be examined. The potential of these levels of temporal and spatial resolution for the study of dynamic materials processes will also be discussed.     

A Model-Based Method for an Online Diagnostic Knowledge-Based System

Fault diagnosis is very important for modern production technology and has received increasing theoretical and practical attention during the last few years. This paper presents a model-based diagnostic method for industrial systems. An online, real-time, deep knowledge based fault detection system has been developed by combining different development environments and tools. The system diagnoses, predicts and compensates faults by coupling symbolic and numerical data in a new environment suitable for the interaction of different sources of knowledge and has been successfully implemented and tested on a real hydraulic system.     

A proteome analysis of the dorsolateral prefrontal cortex in human alcoholic patients

Alcoholic patients commonly experience cognitive decline. It is postulated that cognitive dysfunction is caused by an alcohol-induced region-selective brain damage, particularly to the prefrontal region, and grey and white matter may be affected differently. We used a proteomics-based approach to compare protein expression profiles of the dorsolateral prefrontal cortex (Brodmann area 9 (BA9)) from human alcoholic and healthy control brains. Changes in the relative expression of 110 protein 'spots' were identified in the BA9 grey matter, of which 54 were identified as 44 different proteins. In our recent article, 60 protein spots were differentially expressed in the BA9 white matter and 18 of these were identified (Alexander-Kaufman, K., James, G., Sheedy, D., Harper, C., Matsumoto, I., Mol. Psychiatry 2006, 11, 56-65). Additional BA9 white matter proteins are identified here and discussed in conjunction to our grey matter results. Thiamine-dependent enzymes transketolase and pyruvate dehydrogenase (E1β ubunit) were among the proteins identified. To our knowledge, this is the first time a disruption in thiamine-dependent enzymes has been demonstrated in the brains of 'neurologically uncomplicated' alcoholics. By identifying protein expression changes in prefrontal grey and white matter separately, hypotheses may draw upon more mechanistic explanations as to how alcoholism causes the structural alterations associated with alcohol-related brain damage and cognitive dysfunction.     

LEARNING AND VERIFYING SAFETY CONSTRAINTS FOR PLANNERS IN A KNOWLEDGE‐IMPOVERISHED SYSTEM

We present a Bayesian approach to learning flexible safety constraints and subsequently verifying whether plans satisfy these constraints. Our approach, called the Safety Constraint Learner/Checker (SCLC), infers safety constraints from a single expert demonstration trace and minimal background knowledge, and applies these constraints to the solutions proposed by multiple planning agents in an integrated and heterogeneous ensemble. The SCLC calculates how much to blame plan fragments (partial solutions) generated by the individual planning agents. This information is used when composing these fragments into a final overall plan. In particular, fragments whose safety violations exceed a threshold are rejected. This facilitates the generation of safe plans. We have integrated the SCLC within the Generalized Integrated Learning Architecture, which was designed for Defense Advanced Research Projects Agency (DARPA)’s Integrated Learning (IL) program. The main goal of the IL program is to promote the development and success of sophisticated systems that learn to solve challenging real‐world problems based on a simple demonstration by a human expert and exiguous domain knowledge. We present experimental results showing the advantages of the SCLC on two multiagent problem‐solving tasks that were benchmark applications in DARPA’s IL program.