All scientific activity involves some method of observation and some method of recording what is observed. These activities can be carried out in ways that involve little interaction between subject and object, as is the case when a telescope observes a far-away star. At the other end of the scale are experiments in modern high energy physics in which there is little distinction between the observer and the observed, and the process of observation materially affects the data that are recorded. (...) In this regard, research on human phenomena resembles modern physics more than it does classical astronomy.Research on human phenomena, however, differs from modern physics in the way in which it affects that which is observed. Both the procedures and the findings of research on human phenomena alter the modes of thinking and the self-awareness of the objects of study. (shrink)

Background Sharing anonymized/de-identified clinical trial data and publishing research outcomes in scientific journals, or presenting them at conferences, is key to data-driven scientific exchange. However, when data from scientific publications are linked to other publicly available personal information, the risk of reidentification of trial participants increases, raising privacy concerns. Therefore, we defined a set of criteria allowing us to determine and minimize the risk of data reidentification. We also implemented a review process at Takeda for clinical publications prior to submission (...) for publication in journals or presentation at medical conferences. Methods Abstracts, manuscripts, posters, and oral presentations containing study participant information were reviewed and the potential impact on study participant privacy was assessed. Our focus was on direct and indirect identifiers, such as sex, age or geographical indicators in rare disease studies or studies with small sample size treatment groups. Risk minimization was sought using a generalized presentation of identifier-relevant information and decision-making on data sharing for further research. Additional risk identification was performed based on study participant/personnel parameters present in materials destined for the public domain. The potential for participant/personnel identification was then calculated to facilitate presentation of meaningful but de-identified information. Results The potential for reidentification was calculated using a risk ratio of the exposed versus available individuals, with a value above the threshold of 0.09 deemed an unacceptable level of reidentification risk. We found that in 13% of Takeda clinical trial publications reviewed, either individuals could potentially be reidentified or inappropriate data sharing plans could pose a data privacy risk to study participants. In 1/110 abstracts, 58/275 manuscripts, 5/87 posters and 3/58 presentations, changes were necessary due to data privacy concerns/rules. Despite the implementation of risk-minimization measures prior to release, direct and indirect identifiers were found in 11% and 34% of the analysed documents, respectively. Conclusions Risk minimization using de-identification of clinical trial data presented in scientific publications and controlled data sharing conditions improved privacy protection for study participants. Our results also suggest that additional safeguards should be implemented to ensure that higher data privacy standards are met. (shrink)

Few things seem more a part of the material world than biological specimens. Yet the processes by which collections of specimens are assembled, translated into information, combined with more information, and distributed are taking research repositories into the virtual realm.The term “virtual” has a number of meanings, and so a research repository can qualify as virtual in a variety of ways. The term would seem to apply, for example, to constructing a repository by forming a network among institutions; using the (...) Internet or the World Wide Web to solicit specimens and information; integrating web-based technology into the operation of the bank; using the Internet or web-based technology to manage relationships with donors or collection sites and recipients; and digitizing specimens. The all-digital repository would seem the most virtual of all possible repositories, a true cyberbank. (shrink)

Few things seem more a part of the material world than biological specimens. Yet the processes by which collections of specimens are assembled, translated into information, combined with more information, and distributed are taking research repositories into the virtual realm.The term “virtual” has a number of meanings, and so a research repository can qualify as virtual in a variety of ways. The term would seem to apply, for example, to constructing a repository by forming a network among institutions; using the (...) Internet or the World Wide Web to solicit specimens and information; integrating web-based technology into the operation of the bank; using the Internet or web-based technology to manage relationships with donors or collection sites and recipients; and digitizing specimens. The all-digital repository would seem the most virtual of all possible repositories, a true cyberbank. (shrink)

Many researchers have argued that humanity will create artificial general intelligence (AGI) within the next twenty to one hundred years. It has been suggested that AGI may inflict serious damage to human well-being on a global scale ('catastrophic risk'). After summarizing the arguments for why AGI may pose such a risk, we review the fieldʼs proposed responses to AGI risk. We consider societal proposals, proposals for external constraints on AGI behaviors and proposals for creating AGIs that are safe due to (...) their internal design. (shrink)

As we increasingly delegate decision-making to algorithms, whether directly or indirectly, important questions emerge in circumstances where those decisions have direct consequences for individual rights and personal opportunities, as well as for the collective good. A key problem for policymakers is that the social implications of these new methods can only be grasped if there is an adequate comprehension of their general technical underpinnings. The discussion here focuses primarily on the case of enforcement decisions in the criminal justice system, but (...) draws on similar situations emerging from other algorithms utilised in controlling access to opportunities, to explain how machine learning works and, as a result, how decisions are made by modern intelligent algorithms or 'classifiers'. It examines the key aspects of the performance of classifiers, including how classifiers learn, the fact that they operate on the basis of correlation rather than causation, and that the term 'bias' in machine learning has a different meaning to common usage. An example of a real world 'classifier', the Harm Assessment Risk Tool (HART), is examined, through identification of its technical features: the classification method, the training data and the test data, the features and the labels, validation and performance measures. Four normative benchmarks are then considered by reference to HART: (a) prediction accuracy (b) fairness and equality before the law (c) transparency and accountability (d) informational privacy and freedom of expression, in order to demonstrate how its technical features have important normative dimensions that bear directly on the extent to which the system can be regarded as a viable and legitimate support for, or even alternative to, existing human decision-makers. (shrink)