Competition and cooperation in artificial intelligence standard setting: Explaining emergent patterns

Efforts to set standards for artificial intelligence (AI) reveal striking patterns: technical experts hailing from geopolitical rivals, such as the United States and China, readily collaborate on technical AI standards within transnational standard-setting organizations, whereas governments are much less willing to collaborate on global ethical AI standards within international organizations. Whether competition or cooperation prevails can be explained by three variables: the actors that make up the membership of the standard-setting organization, the issues on which the organization's standard-setting efforts focus, and the “games” actors play when trying to set standards within a particular type of organization. A preliminary empirical analysis provides support for the contention that actors, issues, and games affect the prospects for cooperation on global AI standards. It matters because shared standards are vital for achieving truly global frameworks for the governance of AI. Such global frameworks, in turn, lower transaction costs and the probability that the world will witness the emergence of AI systems that threaten human rights and fundamental freedoms.     

Negligent homicide caused by exhaust gas escaping from a manipulated chimney

A chimney built and operated according to the instructions is supposed to ensure that the combustion gases coming from the fireplace can escape safely. If the operational reliability is impaired, this presents a risk of acute poisoning. The report deals with a negligently caused carbon monoxide poisoning of a married couple as a consequence of improper installation of a cover of the chimney opening. Various causes of fatal poisoning due to defective exhaust systems are discussed in connection with the presented case report.     

The Influence of the Relationship and Motivation on Inequity Aversion in Dogs

One crucial element for the evolution of cooperation may be the sensitivity to others?? efforts and pay-offs in comparison with one??s own costs and gains. If an individual responds to a disadvantageous reward distribution, then it would likely increase its relative fitness compared with those who do not (Brosnan & de Waal, Nature 428:140, 2004). Recent experimental research indicates that sensitivity toward unequal reward distribution is not a uniquely human trait; non-human primates react to inequity if they witness a conspecific that obtains a more attractive reward for the same effort. However, primates are not the only species that cooperate and thus would benefit from inequity aversion. Canids also frequently cooperate with group members (e.g., wolves, wild dogs, etc.) or humans (domestic dogs), allowing for the possibility that they also are sensitive to the reward distribution. We report the findings of two studies of dogs?? responses to inequity and the social factors that mediate such responses. In the first study, we investigated whether domestic dogs showed a response to the inequity of rewards received for the same action in pairs of familiar dogs. We found that dogs showed significant behavioral differences when tested without food rewards when in the presence of a rewarded partner compared to the baseline and asocial control situations. This indicated that the presence of a rewarded partner matters (Range, Horn, Virányi, & Huber, Proc Natl Acad Sci USA 106(1):340?C345, 2009). In contrast to primates, dogs did not react to differences in the quality of food or effort. In the second study, we analyzed whether individual (motivational or personality) characteristics determined the response of each subject to unequal reward distributions or whether the subject??s responses depended on the specific relationship they had with its partner. We found that individual motivation and the quality of the relationship influenced the dogs?? performance in test conditions, but that these factors varied across the control and reward inequity conditions. Overall, our results suggest that inequity aversion in dogs is conditional on their and their partner??s rewards. Whether inequity aversion is based on the same mechanisms in both humans and non-human primates is still unclear.     

Valuations – or How to Say the Unsayable

In this paper, the author revisits the emotive theory of value and argues that values are not entities but nothing other than linguistic fictions. Accordingly, valuations—i.e., valuing actions—can be defined as approving or disapproving attitudes of a subject to some object. In this perspective, values cannot be true or false: What we can do is just compare them with regard to strength. As a consequence, value judgments are to be understood as sentences which are used either to say that a subject s values an object o positively or negatively, or to express (evince) a valuation. The author then shows some relations between normative and evaluative discourses. First, he claims that norms as well as valuations are not true or false. Second, he argues that both may be explained or justified, even if the former are usually justified teleologically whereas the latter are explained referring back to the subject's background and life-style. Third, he notes that a legal order originates from the fact that valuations crystallize into norms. Finally, the author examines some further questions related to his analysis. In particular, he argues that the different realms of values, e.g., morals, aesthetics, politics, etc., do not correspond to different evaluative attitudes, but to different phenomena and diverse spheres of human life.     

Towards a new quality of cooperation? The EU,China, and Central Asian Security in a Multipolar Age

Asia Europe Journal - Against the background of ever stronger linkages between the EU and China as well as their emphatically voiced objective to jointly shape the increasingly multi-polar world,...     

Identifying Counterfeit Cigarettes Using Environmental Pollen Analysis: An Improved Procedure*

Traditional pollen preparation techniques provide clear residues for pollen identification; however, such methods are time-consuming, requiring repeated centrifugation, heating, and digestion with high-concentration hazardous chemicals. Tobacco leaves can effectively trap environmental pollen due to hairy surface and terpene-rich exudates. A new tobacco sample processing method was developed by using different extraction chemistry with surfactant. Marlboro Gold cigarettes were employed as model samples for method development. Parameters critical for pollen extraction, which include number of cigarette sticks used, extraction solution, and extraction temperature, were optimized. By using 1% dishwashing detergent to treat three cigarettes at room temperature, the improved method was able to recover sufficient pollen for microscopic analysis in three repeated centrifuge-washing steps and omit hazardous chemicals involved in traditional methods. We focused on the pollen of common ragweed (Ambrosia artemisiifolia), a plant native to North America, as an indicator to differentiate genuine and counterfeit U.S. brand cigarettes. Results from analyzing randomly purchased genuine (authenticated by forensic examination) and known counterfeit Marlboro Gold provided by law enforcement revealed that a significant amount (39%) of Ambrosia were consistently present in all genuine samples, while counterfeit contained none or only trace count. Similar results were found in other counterfeit U.S. brand cigarettes (all seized in the U.S.) involved in this study as well. Lack of Ambrosia in cigarette strongly indicates the product was not originated in the United States.     

The problem of single parent/child paternity analysis--practical results involving 336 children and 348 unrelated men

In a certain amount of paternity investigations, only DNA from child and alleged father is analyzed, thus increasing the possibility of false paternity inclusions. The aim of this study was to determine how many wrong paternity inclusions could be detected in a rather small geographical area comparing empirical results from 336 children and 348 men (13-15 STRs were investigated per person). This comparison between each child and all unrelated men (i.e. all putative fathers from the other cases) with an especially designed computer program resulted in 116,004 man/child pairs. Less than three excluding STRs were found in 1666 child/unrelated man pairs (1.44% of the comparisons). At least one unrelated man with only two or less STR mismatches could be determined for 322 children (95.8% of all investigated children). In 26 comparisons no STR mismatches between a child and an unrelated man were detected, thus at least one and up to three "second father(s)" under 350 men could be found for 23 children, if the mother is excluded. Paternity probabilities between 95.475% and 99.996% were calculated. Our results underline the difficulties in motherless paternity cases using only STR analysis and advise great precaution in assigning verbal predicates such as "paternity proven" in those investigations.     

The EDNAP mitochondrial DNA population database (EMPOP) collaborative exercises: organisation, results and perspectives

This paper presents an overview of the organisation and the results of the collaborative exercises (CE) of the European DNA Profiling (EDNAP) Group's mitochondrial DNA population database project (EMPOP). The aim of the collaborative exercises was to determine whether uniformity of mtDNA sequencing results could be achieved among different laboratories. These were asked to sequence either the complete mtDNA control region or the two hypervariable regions HVI (16024-16365) and HVII (73-340) from DNA extracts, buccal swabs or bloodstains, proceeding in accordance with the protocol and strategies used in each individual laboratory. The results of the collaborative exercises were employed to identify possible sources of errors that could arise during the analysis and interpretation of mtDNA profiles. These findings were taken as a basis to tentatively make suitable arrangements for the construction of a high quality mtDNA database. One hundred fifty mtDNA profiles were submitted to the evaluating laboratory, and disaccording profiles were classified into four groups corresponding to the source of error: clerical errors, sample mix-ups, contaminations and discrepancies with respect to the mtDNA nomenclature. Overall, 14 disaccording haplotypes (16 individual errors) were observed. The errors included 10 clerical errors, 3 interpretation problems, 2 cases of sample mix-up and 1 case of point heteroplasmic mixture, where the 2 sequencing reactions brought inconsistent base calls. This corresponds to an error rate of 10.7% in a virtual mtDNA database consisting of the collaborative exercise results. However, this estimate is still conservative compared to conclusions drawn by authors of meanwhile numerous publications critically reviewing published mtDNA population databases. Our results and earlier published concerns strongly emphasize the need for appropriate safety regulations when mtDNA profiles are compiled for database purposes in order to accomplish the high standard required for mtDNA databases that are used in the forensic context.