Learning to Live with Strange Error: Beyond Trustworthiness in Artificial Intelligence Ethics

Position papers on artificial intelligence (AI) ethics are often framed as attempts to work out technical and regulatory strategies for attaining what is commonly called trustworthy AI. In such papers, the technical and regulatory strategies are frequently analyzed in detail, but the concept of trustworthy AI is not. As a result, it remains unclear. This paper lays out a variety of possible interpretations of the concept and concludes that none of them is appropriate. The central problem is that, by framing the ethics of AI in terms of trustworthiness, we reinforce unjustified anthropocentric assumptions that stand in the way of clear analysis. Furthermore, even if we insist on a purely epistemic interpretation of the concept, according to which trustworthiness just means measurable reliability, it turns out that the analysis will, nevertheless, suffer from a subtle form of anthropocentrism. The paper goes on to develop the concept of strange error, which serves both to sharpen the initial diagnosis of the inadequacy of trustworthy AI and to articulate the novel epistemological situation created by the use of AI. The paper concludes with a discussion of how strange error puts pressure on standard practices of assessing moral culpability, particularly in the context of medicine.

Ethical Implications of e-Health Applications in Early Preventive Healthcare

As a means of preventive medicine early detection and prevention examinations can identify and treat possible health disorders or abnormalities from an early age onwards. However, pediatric examinations are often widely spaced, and thus only snapshots of the children’s and adolescents’ developments are obtained. With e-health applications parents and adolescents could record developmental parameters much more frequently and regularly and transmit data directly for ongoing evaluation. AI technologies could be used to search for new and previously unknown patterns. Although e-health applications could improve preventive healthcare, there are serious concerns about the unlimited use of big data in medicine. Such concerns range from general skepticism about big data in medicine to specific challenges and risks in certain medical areas. In this paper, we will focus on preventive health care in pediatrics and explore ethical implications of e-health applications. Specifically, we will address opportunities and risks of app-based data collection and AI-based data evaluation for complementing established early detection and prevention examinations. To this end, we will explore the principle of the best interest of the child. Furthermore, we shall argue that difficult trade-offs need to be made between group benefit on the one hand and individual autonomy and privacy on the other.

Moral ambivalence. A comment on non-invasive prenatal testing from an ethical perspective

BACKGROUND

Non-invasive prenatal testing (NIPT) has been available for almost 10 years. In many countries the test attracted considerable criticism from the start. While most critical comments in this context deal with the (alleged) problem of eugenic selection, I will concentrate on a somewhat broader issue.

CONTENT

I will argue that NIPT clearly has the potential to increase reproductive autonomy and benefit expectant parents. However, NIPT can also put people in a situation that is morally overwhelming for them and from which there is no easy way out. In this sense, such tests can have a dilemma-generating effect.

SUMMARY AND OUTLOOK

I will conclude that this can be adequately described by the term "moral ambivalence".

Advance research directives: avoiding double standards

Background

Advance research directives (ARD) have been suggested as a means by which to facilitate research with incapacitated subjects, in particular in the context of dementia research. However, established disclosure requirements for study participation raise an ethical problem for the application of ARDs: While regular consent procedures call for detailed information on a specific study (“token disclosure”), ARDs can typically only include generic information (“type disclosure”). The introduction of ARDs could thus establish a double standard in the sense that within the context of ARDs, type disclosure would be considered sufficient, while beyond this context, token disclosure would remain necessary.

Main body

This paper provides an ethical analysis of ARDs, taking into account the results of numerous empirical studies that have been performed so far. It will be argued that a revised understanding of informed consent can allow for context-sensitive disclosure standards. As a consequence, ARDs that include type disclosure can be acceptable under suitable circumstances. Such an approach raises a number of objections. A thorough examination shows, however, that they are not sufficient to justify a rejection of the approach.

Conclusion

The approach presented in this paper avoids introducing a double standard. It is, therefore, more suitable for the implementation of ARDs than established approaches.

[The predictable human : Possibilities and risks of AI-based prediction of cognitive abilities, personality traits and mental illnesses]

New approaches to the use of artificial intelligence (AI) to analyze data from neuroimaging but also passively collected data from so-called wearables, such as smartphones or smartwatches, as well as data that can be extracted from social media and other online activities, already make it possible to predict cognitive abilities, personality traits, and mental illnesses, as well as to reveal acute mental states. In this article, we explain the methodological concepts behind these current developments, illuminate the possibilities and limitations, and address ethical and social aspects arising from the use.

Cannabis-Associated Psychotic-like Experiences Are Mediated by Developmental Changes in the Parahippocampal Gyrus

OBJECTIVE

Cannabis consumption during adolescence has been reported as a risk factor for psychotic-like experiences (PLEs) and schizophrenia. However, brain developmental processes associated with cannabis-related PLEs are still poorly described.

METHOD

A total of 706 adolescents from the general population who were recruited by the IMAGEN consortium had structural magnetic resonance imaging scans at both 14 and 19 years of age. We used deformation-based morphometry to map voxelwise brain changes between the two time points, using the pairwise algorithm in SPM12b. We used an a priori region-of-interest approach focusing on the hippocampus/parahippocampus to perform voxelwise linear regressions. Lifetime cannabis consumption was assessed using the European School Survey Project on Alcohol and other Drugs (ESPAD), and PLEs were assessed with the Comprehensive Assessment Psychotic-like experiences (CAPE) tool. We first tested whether hippocampus/parahippocampus development was associated with PLEs. Then we formulated and tested an a priori simple mediation model in which uncus development mediates the association between lifetime cannabis consumption and PLEs.

RESULTS

We found that PLEs were associated with reduced expansion within a specific region of the right hippocampus/parahippocampus formation, the uncus (p = .002 at the cluster level, p = .018 at the peak level). The partial simple mediation model revealed a significant total effect from lifetime cannabis consumption to PLEs (b = 0.069, 95% CI = 0.04-0.1, p =2 × 10^-16), as well as a small yet significant, indirect effect of right uncus development (0.004; 95% CI = 0.0004-0.01, p = .026).

CONCLUSION

We show here that the uncus development is involved in the cerebral basis of PLEs in a population-based sample of healthy adolescents.

Your evidence? Machine learning algorithms for medical diagnosis and prediction

Computer systems for medical diagnosis based on machine learning are not mere science fiction. Despite undisputed potential benefits, such systems may also raise problems. Two (interconnected) issues are particularly significant from an ethical point of view: The first issue is that epistemic opacity is at odds with a common desire for understanding and potentially undermines information rights. The second (related) issue concerns the assignment of responsibility in cases of failure. The core of the two issues seems to be that understanding and responsibility are concepts that are intrinsically tied to the discursive practice of giving and asking for reasons. The challenge is to find ways to make the outcomes of machine learning algorithms compatible with our discursive practice. This comes down to the claim that we should try to integrate discursive elements into machine learning algorithms. Under the title of "explainable AI" initiatives heading in this direction are already under way. Extensive research in this field is needed for finding adequate solutions.

Neural Correlates of Failed Inhibitory Control as an Early Marker of Disordered Eating in Adolescents

BACKGROUND

Binge eating and other forms of disordered eating behavior (DEB) are associated with failed inhibitory control. This study investigated the neural correlates of failed inhibitory control as a potential biomarker for DEB.

METHODS

The study used prospective longitudinal data from the European IMAGEN study adolescent cohort. Participants completed baseline assessments (questionnaires and a brain scan [functional magnetic resonance imaging]) at 14 years of age and a follow-up assessment (questionnaires) at 16 years of age. Self-reported binge eating and/or purging were used to indicate presence of DEB. Neural correlates of failed inhibition were assessed using the stop signal task. Participants were categorized as healthy control subjects (reported no DEB at both time points), maintainers (reported DEB at both time points), recoverers (reported DEB at baseline only), and developers (reported DEB at follow-up only). Forty-three individuals per group with complete scanning data were matched on gender, age, puberty, and intelligence (N = 172).

RESULTS

At baseline, despite similar task performance, incorrectly responding to stop signals (failed inhibitory control) was associated with greater recruitment of the medial prefrontal cortex and anterior cingulate cortex in the developers compared with healthy control subjects and recoverers.

CONCLUSIONS

Greater recruitment of the medial prefrontal and anterior cingulate regions during failed inhibition accords with abnormal evaluation of errors contributing to DEB development. As this precedes symptom onset and is evident despite normal task performance, neural responses during failed inhibition may be a useful biomarker of vulnerability for DEB. This study highlights the potential value of prospective neuroimaging studies for identifying markers of illness before the emergence of behavior changes.

Working with Research Integrity-Guidance for Research Performing Organisations: The Bonn PRINTEGER Statement

This document presents the Bonn PRINTEGER Consensus Statement: Working with Research Integrity-Guidance for research performing organisations. The aim of the statement is to complement existing instruments by focusing specifically on institutional responsibilities for strengthening integrity. It takes into account the daily challenges and organisational contexts of most researchers. The statement intends to make research integrity challenges recognisable from the work-floor perspective, providing concrete advice on organisational measures to strengthen integrity. The statement, which was concluded February 7th 2018, provides guidance on the following key issues: § 1. Providing information about research integrity § 2. Providing education, training and mentoring § 3. Strengthening a research integrity culture § 4. Facilitating open dialogue § 5. Wise incentive management § 6. Implementing quality assurance procedures § 7. Improving the work environment and work satisfaction § 8. Increasing transparency of misconduct cases § 9. Opening up research § 10. Implementing safe and effective whistle-blowing channels § 11. Protecting the alleged perpetrators § 12. Establishing a research integrity committee and appointing an ombudsperson § 13. Making explicit the applicable standards for research integrity.

2006 GEP-ISFG collaborative exercise on mtDNA: reflections about interpretation, artefacts, and DNA mixtures

We report the results of the seventh edition of the GEP-ISFG mitochondrial DNA (mtDNA) collaborative exercise. The samples submitted to the participant laboratories were blood stains from a maternity case and simulated forensic samples, including a case of mixture. The success rate for the blood stains was moderate ( approximately 77%); even though four inexperienced laboratories concentrated about one-third of the total errors. A similar success was obtained for the analysis of mixed samples (78.8% for a hair-saliva mixture and 69.2% for a saliva-saliva mixture). Two laboratories also dissected the haplotypes contributing to the saliva-saliva mixture. Most of the errors were due to reading problems and misinterpretation of electropherograms, demonstrating once more that the lack of a solid devised experimental approach is the main cause of error in mtDNA testing.

A comparative analysis of selected European guidelines and recommendations for biobanks with special regard to the research / non-research distinction

Classifying an activity as research has considerable normative consequences: On the one side research activities are especially protected ("freedom of research"), on the other side stricter ethical rules legal regulations may apply to such activities, e.g. compulsory review by ethics committees. It is therefore important to find criteria to delimit research from non-research activities. The aim of the present study is to analyse some European guidelines and recommendations for biobanks and to see whether these documents provide comments that help to elucidate the research/non-research distinction.

Delimiting the concept of research: an ethical perspective

It is important to be able to offer an account of which activities count as scientific research, given our current interest in promoting research as a means to benefit humankind and in ethically regulating it. We attempt to offer such an account, arguing that we need to consider both the procedural and functional dimensions of an activity before we can establish whether it is a genuine instance of scientific research. By placing research in a broader schema of activities, the similarities and differences between research activities and other activities become visible. It is also easier to show why some activities that do not count as research can sometimes be confused with research and why some other activities can be regarded only partially as research. Although the concept of research is important to delimit a class of activities which we might be morally obliged to promote, we observe that the class of activities which are regarded as subject to ethical regulation is not exhausted by research activities. We argue that, whether they be research or not, all the activities that are likely to affect the rights and interests of the individuals involved and impact on the rights and interests of other individuals raise ethical issues and might be in need of ethical regulation.

What should we want to know about our future? A Kantian view on predictive genetic testing

Recent advances in genomic research have led to the development of new diagnostic tools, including tests which make it possible to predict the future occurrence of monogenetic diseases (e.g. Chorea Huntington) or to determine increased susceptibilities to the future development of more complex diseases (e.g. breast cancer). The use of such tests raises a number of ethical, legal and social issues which are usually discussed in terms of rights. However, in the context of predictive genetic tests a key question arises which lies beyond the concept of rights, namely, What should we want to know about our future? In the following I shall discuss this question against the background of Kant's Doctrine of Virtue. It will be demonstrated that the system of duties of virtue that Kant elaborates in the second part of his Metaphysics of Morals offers a theoretical framework for addressing the question of a proper scope of future knowledge as provided by genetic tests. This approach can serve as a source of moral guidance complementary to a justice perspective. It does, however, not rest on the-rather problematic--claim to be able to define what the "good life" is.

Identification of missing persons: the Spanish "Phoenix" program

In 1999, Spain was the first country to officially start a National Program to try to identify cadavers and human remains which could not be identified by the use of traditional forensic approaches. This attempt is called "Phoenix Program". Two independent mitochondrial DNA (mtDNA) databases were generated, which can automatically compare and match identical or similar sequences. One is the Reference Database, with mtDNA sequences from maternal relatives of missing persons, who provide the samples (buccal swabs) voluntarily; the other is the Questioned Database, comprised of mtDNA data of unknown remains and cadavers. Although the first phase of the program (typing of all unidentified human remains) will probably not be completed until December 2003, positive identifications are being made in the interim. To date, more than 1,200 families have contacted Phoenix, and at least 280 reference samples and 48 questioned evidences have been analyzed. When mtDNA matches are found, another independent analysis is performed as a part of the quality control mechanism. Once a match is confirmed (so far in 6 cases), an attempt is made to analyze short tandem repeat (STR) loci. We call for international collaboration to make this effort valuable worldwide.

Fractal Characterization of Wide Pore Range Catalysts: Application to Pd-Ag/SiO(2) Xerogels

Fractal analysis has been applied to characterize the structure of Pd-Ag/SiO(2) catalysts dried under vacuum (150 degrees C and 12 hPa) with different concentrations of Pd-Ag. Nitrogen adsorption-desorption, mercury porosimetry, and small-angle X-ray scattering measurements have been used. The different approaches to fractal analysis and their conditions of applicability are briefly described. Results are discussed in relation to those derived previously from classical interpretations. This analysis shows that Pd-Ag/SiO(2) xerogels exhibit a very open self-similar pore structure analogous to the structure of supercritically dried aerogels and that the micropore texture can be tailored by varying the silver content. Copyright 2000 Academic Press.

High-dose granulocyte-colony stimulating factor (G-CSF) in vitro induces the growth of high proliferative potential colony forming cells (HPP-CFC) in patients undergoing blood stem cell mobilization

We evaluated the role of high-dose granulocyte colony stimulating factor (G-CSF) in vitro, in inducing the generation of high-proliferative potential colony forming cells (HPP-CFC), from either mononuclear cells or purified CD34+ cells. Both normal controls and patients undergoing peripheral blood stem cell (PBSC) mobilization and transplantation were studied. In serum-driven agar cultures, G-CSF stimulated the proliferation of HPP-CFC in a dose dependent manner (r = 0.92). The number of HPP-CFC was four-fold greater in mobilized patients than in normal controls. Purified CD34+ cells yielded 11-fold more colonies than mononuclear cells. HPP-CFC from mobilized patients showed replating capacity, giving rise to secondary colonies of more mature appearance. In serum-free cultures, the effect of G-CSF appeared to be mediated by synergistic interaction with stem cell factor. Our results suggest that G-CSF stimulates primitive hematopoietic cells that are detectable in increased amounts in patients receiving mobilization therapy. Therefore, determination of G-CSF induced HPP-CFC could be a useful tool in the evaluation of mobilization strategies.

Juvenile chronic lymphocytic leukemia with unusual intracisternal inclusions: an ultrastructural study

The electron microscopic analysis of intracisternal inclusions in lymphocytes of the bone marrow and peripheral blood in a case of juvenile chronic lymphocytic leukemia is described. These inclusions consist of well-ordered microtubules attached to a central axis. The contribution of electron microscopic analysis in establishing the substructural pattern of these inclusions is discussed.

[Chediak-Higashi-like inclusions in acute myeloblastic leukemia. Ultrastructural study]

The pseudo-Chédiak-Higashi anomaly is characterized by the presence of giant granules in the cytoplasm of blast cells in acute leukemia. We report here a new case of acute myelogenous leukemia (M2 type) with this alteration. The granules were azurophilic or eosinophilic and reacted strongly to peroxidase stain. Ultrastructural studies showed that the granules contained a dense matrix and occasionally "finger print" structures at the periphery; in some inclusions, fibrillar structures of myelinic figures could be seen. The matrix was reactive to peroxidase and small vesicles were prominent in the cytoplasm near the granules. We conclude that the giant granules could have been formed by the fusion of primary granules and/or by the fusion of these small dense vesicles.

[Pure red-cell aplasia associated with chronic myeloproliferative syndromes]

Pure red cell aplasia is manifested by unregenerative anemia with intense erythroid hypoplasia in the bone marrow and absolute reticulocytopenia. Its association with chronic myeloproliferative syndrome (CMS) is infrequent. The appearance of both processes may or may not be simultaneous and accidental association cannot be ruled out. Two cases of pure red cell aplasia were studied; one associated to Philadelphia positive chronic chromosome myeloid leukemia and the other associated to atypical CMS. In the first case no response was obtained with prednisone alone or associated to cyclophosphamide. In the second patient treatment with danazol was not efficient, however, corticotherapy produced a positive and maintained response. Treatment of pure red cell aplasia associated to CMS must be the same as that employed when this disease is presented as isolated or associated to other processes.

The fine structural localization of endogenous and exogenous peroxidase activity in human bone marrow mast cells under pathological conditions

We have examined the ultrastructural characteristics of peroxidase activity in human bone marrow mast cells. These studies were performed in three patients with systemic mast cell disease, and in another six patients showing bone marrow mast cell hyperplasia. Endogenous peroxidase activity was localized in the perinuclear cisternae and strands of endoplasmic reticulum, but never in the granules. We have also demonstrated the "in vivo" existence of exogenous peroxidase activity in two of the three cases of systemic mast cell disease. The peroxidase internalization involved its binding to the plasma membrane, followed by its incorporation into the cell by a general endocytic process comprising the uptake of dispersed peroxidase-positive material mainly by phagocytosis of granular structures containing peroxidase. The exogenous peroxidase appeared in non-membrane bound granules, vacuoles or aggregates, but we have never seen the enzyme linked to the mast cell granules.

Unilateral adrenal hemorrhage in forty-five-year-old woman

We herein describe a case of adrenal hemorrhage on the right side in a forty-five-year-old woman. Her medical history did not show any evidence of an acute bleeding. The hematoma was removed and adrenalectomy performed. A veinectasis and a localized vasculitis are discussed as causes of bleeding.

[Complication of ureterorenoscopy]

We describe failures and complications of ureterorenoscopy based on our own experience and the literature. There is usually no consequence when perforating the ureter--which happens in 10 per cent of the cases--if a splint or a percutaneous nephrostomy is applied. Ureteral rupture makes an open intervention necessary, but happens rarely. Since complications of ureterorenoscopy are rare and the rate of success very high, open operations on the ureter should become less frequent.

[Complications of percutaneous kidney surgery--prevention and control]

Possibilities and types of complications, how to avoid and deal with them, are outlined in our material of 43 exactly defined cases. Extravasation and perforation of the renal calyces, the leaving behind of small concrements, perforation of neighbouring organs, hemorrhage, infection and late complications are discussed. Other published series, are analyzed (about 4000 patients up to now). However, often no information about incidence and type of complications is given. One must assume that several complications are not mentioned depending upon the way each author judges the complications. In general mild complications like extravasation of irrigation water and/or contrast medium and even hemorrhage diminish with practise. Inspite of our small material of about 50 percutaneous operations on the renal calyces a critical view of the larger series in the literature concerning complications is necessary. The most important points are avoiding and dealing with the discussed complications. Since percutaneous surgery on the kidney does not utilize a natural channel like the other endoscopic procedures, but creates its own channel, one should call this method a "half-open" operation.