Recently, artificial intelligence and machine learning in general have demonstrated remarkable performances in many tasks, from image processing to natural language processing, especially with the advent of deep learning (DL). Along with research progress, they have encroached upon many different fields and disciplines. Some of them require high level of accountability and thus transparency, for example, the medical sector. Explanations for machine decisions and predictions are thus needed to justify their reliability. This requires greater interpretability, which often means we need to understand the mechanism underlying the algorithms. Unfortunately, the blackbox nature of the DL is still unresolved, and many machine decisions are still poorly understood. We provide a review on interpretabilities suggested by different research works and categorize them. The different categories show different dimensions in interpretability research, from approaches that provide “obviously” interpretable information to the studies of complex patterns. By applying the same categorization to interpretability in medical research, it is hoped that: 1) clinicians and practitioners can subsequently approach these methods with caution; 2) insight into interpretability will be born with more considerations for medical practices; and 3) initiatives to push forward data-based, mathematically grounded, and technically grounded medical education are encouraged.

A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI

Gru nd legung zur Metaphysik der Sitten

What are scoping studies? A review of the nursing literature

ionist strategies, 24 commitments, 181–182; normative, Accidentally true beliefs, 114 187–189, 197 Acknowledgments, 173–175, 177–179, 181 Assimilation, 2–3, 11 Acquired beliefs, 106–107, 121 Assimilationalism, 34 Action, 15, 24, 37–38, 79–96. See also Assimilation theories, 3, 6 Commitment(s); Practical Associationist strategies, 24 reasoningM Asymmetric substitution inferences, Adverbs, 206n14 135–136, 142–143, 144–146, 149, 151–154 Analysis of meanings, 31. See also Atomism, 15–16, 29, 31 Concepts Attributions, 168–169, 173–175, 178, Anscombe, G. E. M., 82 180–181 Articulation: inferential, 17, 22, 26, 28, Authority, 11, 31, 44, 76, 165 39, 44, 47, 52, 63, 108, 129, 161–166, 173, Awareness, 2, 7, 16–17, 20, 22, 24–25, 31, 178, 183, 193, 195, 220n4; conceptual, 35, 48, 157, 160 32; logical, 52; social, 158–159, 163 Ascriptions, 175–178, 183 Belief (s), 4–6, 12, 16–17, 37, 43, 53, 69, Assertibilism, 64, 185–189 82–84, 187, 210n2, 211n6; true, 39, Assertibility conditions, 196–198, 97–101, 106–108, 111–112, 114, 118–119; 201–202 and reliabilism, 97–121; acquired, Assertional commitments, 179 106–107, 121; perceptual, 108–109, 121; Assertional contents, 192, 194, 196 and reasons, 158–161, 165–166, 168, Assertional practices, 189, 193, 197 174–177, 180–182, 189, 191–192; Assertion(s): pragmatic nature of, collateral, 219n3. See also Commit11–12, 17, 40; and inferences, 15, 17, 43; ment(s) and singular terms, 125, 129, 145–147, Belnap, Nuel D., Jr., 68 152–153; and reasons, 161, 165, 189–195. Bennett, Jonathan, 70 See also Commitment(s); Language; Blindsight, super, 102–106 Copyright © 2000 The President and Fellows of Harvard College Blindspots, 39, 97–122 Boole, George, 58 Boolean logic, 58–59 Bottom-up semantic explanation, 12–15 Bradley, F. H., 140 Brandom, Robert B.: Making It Explicit, 10, 35–36, 80–81, 119, 205n7, 209n4 Carnap, Rudolph, 7, 23, 118; The Logical Syntax of Language, 50 Causal chains, 113–115. See also Reliabilism Causal functionalism, 95 Causes, 45–46, 53, 84, 94 Certainty, 210n1 Ceteris paribus clauses, 88–89 Charged words, 69–71 Chomsky, Noam, 126–127 Circumstances, 62–66, 69, 72, 74–75, 221n7. See also Assertion(s); Consequences Claims, 43–44, 48, 56–57, 64, 81–82, 86, 196–197; assertion of, 11–12, 125, 194; in semantic explanations, 12–15; expressive nature of, 16–19; conditional, 21–22, 86–87; inferentially articulated, 26, 28–29; knowledge, 39; and objects, 40; conceptual content of, 50; reliable, 107; counter-, 149; and reasons, 158–162, 164–169, 174–175, 180–183, 193; understanding, 189, 191; undertaking, 192; quantificational, 214n13 Classification, conceptual, 17, 48 Cognitive activity, 46, 157, 160 Cognitive responsibility, 105–106 Collateral beliefs, 219n3 Commitment-preserving inferences, 43–44, 195, 201. See also Inference(s) Commitment(s), 31, 73, 220nn5–6, 221n7; inferential, 30, 64, 69–79, 91, ◆ Index 223 147–149, 178; and reasons, 43, 163–168, 174–175, 177–178, 183, 189–196; incompatible, 44; explicit/implicit, 56–57; and practical reasoning, 80–85, 91–96; practical, 93–96; discursive, 94, 164–165; consequential, 108–109, 191–192; and reliabilism, 108–109, 118–120; SMSICs, 136–141; and singular terms, 151–153; substitutional, 178–182; assertional, 179; normative, 189–196, 198–203. See also Action; Assertion(s); Belief(s); Entitlement Committive inferences, 194 Communication: and representation, 165–169, 183; and beliefs, 180 Concepts, 1–22, 80, 108–109, 160 Conceptual articulation, 32 Conceptual assimilationalism, 34 Conceptual Blindspot, 39, 109, 117–118,

Articulating Reasons: An Introduction to Inferentialism

Creating the Kingdom of Ends

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.

https://onlinelibrary.wiley.com/doi/epdf/10.1002/hbm.24886

Your evidence? Machine learning algorithms for medical diagnosis and prediction

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.

/pdf/working-with-research-integrity-guidance-for-research-cc3o2z7byk.pdf

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

Forschung am Menschen: Elemente einer ethischen Theorie biomedizinischer Humanexperimente

In this paper, I examine whether the use of artificial intelligence (AI) and automated decision-making (ADM) aggravates issues of discrimination as has been argued by several authors. For this purpose, I first take up the lively philosophical debate on discrimination and present my own definition of the concept. Equipped with this account, I subsequently review some of the recent literature on the use AI/ADM and discrimination. I explain how my account of discrimination helps to understand that the general claim in view of the aggravation of discrimination is unwarranted. Finally, I argue that the use of AI/ADM can, in fact, increase issues of discrimination, but in a different way than most critics assume: it is due to its epistemic opacity that AI/ADM threatens to undermine our moral deliberation which is essential for reaching a common understanding of what should count as discrimination. As a consequence, it turns out that algorithms may actually help to detect hidden forms of discrimination.

/pdf/discrimination-in-the-age-of-artificial-intelligence-3zoix4l4zi.pdf

Discrimination in the age of artificial intelligence

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.

Bert Heinrichs

Papers

Your evidence? Machine learning algorithms for medical diagnosis and prediction

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

Forschung am Menschen: Elemente einer ethischen Theorie biomedizinischer Humanexperimente

Discrimination in the age of artificial intelligence

Delimiting the concept of research: An ethical perspective