IT and Society Lecture 13: Research and Ethics PDF

IT and Society Lecture 13: Research and Ethics Prof. Jens Grossklags, Ph.D. Professorship of Cyber Trust Department of Computer Science School of Computation, Information and Technology Technical University of Munich July 15, 2024 Recap – Fairness, Accountability & Transparency (FAT) 1. Fairness,...

IT and Society Lecture 13: Research and Ethics Prof. Jens Grossklags, Ph.D. Professorship of Cyber Trust Department of Computer Science School of Computation, Information and Technology Technical University of Munich July 15, 2024 Recap – Fairness, Accountability & Transparency (FAT) 1. Fairness, accountability and transparency can serve as ethical measurements 2. Fairness, accountability and transparency are trust-enhancing factors  product adoption 3. While algorithms outperform humans on a variety of tasks, they may systematically and consistently discriminate if the dataset contains (human) biases 4. Raw data is an oxymoron? 2 Recap – Fairness, Accountability & Transparency (FAT) 1. Algorithmic systems can (in most cases) only implement one conceptualization of fairness 2. The GDPR contains a “Right to Explanation,” but only grants data subjects ex ante explanations – Good when choices can be explained in an exhaustive and easy-to- understand fashion (e.g., two simple options with easy-to-explain consequences) – Otherwise, ex post explanations are likely preferable 3. Likely all ML-based systems face FAT challenges if they make prediction on individuals 3 The Ethical Aspects of Nudging  Should governments nudge us for “good”? Or is nudging morally problematic? E.g., We may know that we will be nudged. But we do not know when we are being nudged. When we are nudged to eat fruit rather than sweets, have we been “tricked”? 4 The Ethical Aspects of Nudging  Three principles for the use of nudges (proposed by Richard Thaler and Cass Sunstein): The nudge should be transparent and never misleading. The nudge should be easily opted-out of. There should be good reason to believe that the behavior being encouraged will improve the welfare of those being nudged. 5 Research & Ethics 6 Plagiarism Original “google paper” by founders Brin and Page duplicated without reference 7 https://doi.org/10.1038/d41586-023-03974-8 8 Focus on Human (Participants) 9 Research Ethics: Two Definitions Research ethics concerns the responsibility of researchers to be honest and respectful to all individuals who are affected by their research studies or their reports of the studies’ results. A set of moral and social standards that includes both prohibitions against and prescriptions for specific kinds of behavior in research. 10  Doctors’ Trials (1946 – 1947) 23 defendants (including 20 medical doctors) Accused of having been involved in: - Cruel and inhumane “medical” experimentation on prisoners and others without consent - Mass murder under the guise of euthanasia 7 were executed in 1948 11 Nuremberg Code Defense argued: – No law separating legal from illegal medical research Response: Memorandum to the United States Counsel for War Crimes outlining points for legitimate medical research – Evolved to a list of 10 points – Research must balance expected benefits against risks to those involved – Likely motivated by Hippocratic Oath Earliest expression of medical ethics “I swear by Apollo Physician, by Asclepius, by Hygieia, by Panacea, and by all the gods and goddesses… “ 12 Hippocratic Oath: Modern Version (U.S. medical schools) I swear to fulfill, to the best of my ability and judgment, this covenant: I will respect the hard-won scientific gains of those physicians in whose steps I walk, and gladly share such knowledge as is mine with those who are to follow. I will apply, for the benefit of the sick, all measures [that] are required, avoiding those twin traps of overtreatment and therapeutic nihilism. I will remember that there is art to medicine as well as science, and that warmth, sympathy, and understanding may outweigh the surgeon's knife or the chemist's drug. I will not be ashamed to say "I know not," nor will I fail to call in my colleagues when the skills of another are needed for a patient's recovery. I will respect the privacy of my patients, for their problems are not disclosed to me that the world may know. Most especially must I tread with care in matters of life and death. If it is given me to save a life, all thanks. But it may also be within my power to take a life; this awesome responsibility must be faced with great humbleness and awareness of my own frailty. Above all, I must not play at God. I will remember that I do not treat a fever chart, a cancerous growth, but a sick human being, whose illness may affect the person's family and economic stability. My responsibility includes these related problems, if I am to care adequately for the sick. I will prevent disease whenever I can, for prevention is preferable to cure. I will remember that I remain a member of society, with special obligations to all my fellow human beings, those sound of mind and body as well as the infirm. If I do not violate this oath, may I enjoy life and art, respected while I live and remembered with affection thereafter. May I always act so as to preserve the finest traditions of my calling and may I long experience the joy of healing those who seek my help. 13 Read it once in full. Nuremberg Code ‒ 10 standards I. The voluntary consent of the subject is absolutely essential. II. The study should yield fruitful results for the good of society, unprocurable by other means. III. The study should be designed and based on results from animal experimentation and knowledge of natural history, such that the Nowadays controversial. anticipated results justify the performance of the experiment. IV. Experiment should be conducted to avoid all unnecessary physical and mental suffering and injury. V. No experiment should be conducted if there is prior reason to believe that death or serious injury could occur. 14 Nuremberg Code ‒ 10 standards VI. The degree of risk should never exceed that determined by the humanitarian importance of the problem to be solved. VII. Proper preparations should be made to protect the subject against injury, disability or death. VIII.The experiment should only be conducted by scientifically qualified persons. IX. The human subject should be at liberty to bring the experiment to an end. X. During the experiment the scientist must bring it to a close if there is probable cause to believe that it could result in injury, disability or death. 15  The End of a History of Serious Misteps? 16 Unfortunately not… An astonishingly sad example: Perhaps the most notorious example of unethical research practice in modern times, impoverished and illiterate, syphilis positive African American men in Tuskegee, Alabama were lied to and misled in the name of science. What happened? [The Tuskegee Syphilis Study (1932 – 1972)] 17 Tuskegee Syphilis Study For 40 years, … …the U.S. Public Health Service conducted the “Study of Untreated Syphilis” on 600 subjects (399 with pre-existing syphilis and 201 without) to study the effects of the disease. As part of the trial, subjects were not afforded the opportunity to provide informed consent to participate, nor were they treated adequately with antibiotics such as Penicillin which became available as a known treatment during the course of the study in the 1940’s. Instead, they were provided a placebo and led to believe they were receiving free health care from the U.S. Government. As a result, many of these men died from Syphilis or complications related to it, infected others with the disease, and passed congenital syphilis on to their children. [The Tuskegee Syphilis Study (1932 – 1972)] 18  More Detailed Case: The Milgram Experiment 19 20 What was the Experiment? The Milgram Experiment was a social psychology experiment conducted by Yale University psychologist Stanley Milgram. – First conducted in 1961 It measured the willingness of participants to obey an authority figure who instructed them to perform acts that possibly conflicted with their personal conscience. – Motivated by (late) Nuremberg trials Stanley Milgram 21 Experiment Participants There were three roles in the experiment: 1. Experimenter (E) 2. Teacher (T) 3. Learner (L) In this experiment, both the experimenter and the learner were actors following a script. The only person really being experimented on was the “teacher.” 22 Experimenter and Learner Advertisement in the paper, offering $4.50 per hour to participate in an experiment regarding “psychology and learning.” Role of the experimenter was played by a stern, impassive biology instructor dressed in a technician’s coat, and the learner was played by an Irish-American accountant trained to act for the role. 23 Setup The participant (i.e., the teacher) and the learner were told by the experimenter that they would be participating in an experiment helping his study of memory and learning in different situations. Two slips of paper were then presented to the participant and to the actor. The participant was led to believe that one of the slips said “learner” and the other said “teacher,” and that he and the actor had been given the slips randomly. In reality, both said “teacher.” 24 Strapping Down the Learner The “teacher” then had to help strap the “learner” into a chair and attach electrical clamps to him. The “learner” mentioned that he had a heart condition. 25 Separation The “teacher” and the experimenter then went to a connected, room. The “teacher” was given a 45-volt electric shock from the electro-shock generator as a sample of the shock that the “learner” would supposedly receive during the experiment. 26 Procedure of the “Test” The “teacher” was then given a list of word pairs which he was to teach the learner. The teacher began by reading the list of word pairs to the learner. The teacher would then read the first word of each pair and read four possible answers. The learner would press a button to indicate his response. If the answer was incorrect, the teacher should administer a shock to the learner, with the voltage increasing for each wrong answer. If correct, the teacher would read the next word pair. 27 Room Setup 28 Deception The subjects (“Teacher”) believed that for each wrong answer, the learner was receiving actual shocks. In reality, there were no shocks. After the confederate was separated from the subject, the confederate set up a tape recorder integrated with the electro- shock generator, which played pre-recorded sounds for each shock level. 29 Impact of Shocks After a number of voltage level increases, the actor started to bang on the wall that separated him from the subject. After several times banging on the wall and complaining about his heart condition, all responses by the learner would cease. 30 Early Reactions At this point, many participants indicated their desire to stop the experiment and to check on the learner. Some test subjects paused at 135 volts and began to question the purpose of the experiment. Participants (in the first experiment) continued after being assured that they would not be held responsible. 31 Small Prompts If at any time the subject indicated his desire to halt the experiment, he was given a succession of verbal prods by the experimenter, in this order: “Please continue” “The experiment requires that you continue” “It is absolutely essential that you continue” “You have no other choice, you must go on” 32 Immediate Impact on Participants Some individuals appeared traumatized while participating – Profuse sweating and trembling – 10% left extremely upset – Some others broke into unexplained hysterical laughter Mentally stable, healthy individuals were emotionally distraught after only 20 minutes – Believed they were causing another human to suffer Observed reactions: – Twitching – Stuttering – Twisting hands – Pulling on earlobes – Nervous laughter 33 Putting a Stop to Things If the subject still wished to stop after all four successive verbal prods, the experiment was halted. Otherwise, it was halted after the subject had given the maximum 450-volt shock three times in succession. 34 Results In Milgram’s first set of experiments, 65 percent (26 of 40) of experiment participants administered the experiment’s final 450-volt shock, though many were very uncomfortable doing so – At some point, every participant paused and questioned the experiment. However, no participant completely refused to administer shocks before 300 volts. 35 Perception vs. Reality During a conference, before going on to reveal what the participants did, Milgram asked how conference attendees would have reacted in such a situation. Unanimously, they replied that they would have stopped despite the experimenter’s orders at less than 300V. I observed a mature and initially poised businessman enter the laboratory smiling and confident. Within 20 minutes he was reduced to a twitching, stuttering wreck, who was rapidly approaching nervous collapse. He constantly pulled on his ear lobe and twisted his hands. At one point he pushed his fist into his forehead and muttered “Oh God, let’s stop it.” And yet he continued to respond to every word of the experimenter, and obeyed to the end.” 36 Theoretical Contributions Theory of Conformism: A subject who has neither ability nor expertise to make decisions, especially in a crisis, will leave decision making to the group and its hierarchy. The group is the person’s behavioral model. Agentic State Theory: The essence of obedience consists in the fact that a person comes to view himself as the instrument for carrying out another person’s wishes, and he therefore no longer sees himself as responsible for his actions. – Very influential study: Changed perception about Nazi-Germany 37 Criticism by Another Psychologist Diana Baumrind (1927 – 2018) Primarily 3 ethical considerations: – Milgram was wrong in deceiving his subjects as to the true purpose of the experiment – As a part of this deception, he was wrong in letting them believe they really were inflicting pain on another human being – Worst of all, he was wrong to put them in a conflict situation that some of them found stressful Additional arguments: – Milgram’s experiment did not provide adequate measures to protect participants from the stress and realization that they were capable of brutal actions – The entire experiment should have been terminated at the first indication of discomfort in the participants – Because of the intensity of the experience, participants might be alienated from future participation in psychological research 38 Your opinion? Milgram’s Defense He argued that adequate measures were taken to protect participants – Participants could withdraw from the study at any time – The deception was explained at the conclusion of the experiment – His subjects volunteered to take part in the experiment – Apart from what appeared to be their inborn urge to obey an ‘authority’ figure, nothing stopped them refusing to comply with the instruction – After the experiment all subjects were fully debriefed and assured that they had not in fact hurt anybody and had nothing to be ashamed of – that their behavior was normal and understandable 39 Milgram’s Defense (2) Deception was necessary as evidenced by the mistaken predictions of the results – Neither Milgram nor any of the psychologists whom he consulted before the investigation believed that these volunteers would behave in the way they did No indications that stress undertaken by participants had any lasting or injurious effects Follow-up questionnaires and interviews, several months and a year after the experiment, showed participants were not alienated from future research – The experience was worthwhile, even positively life-altering for them – Far from criticising Milgram, his subjects maintained that, though stressed, they were glad to have taken part and discovered important aspects of their own psychology 40 Impact on Research Ethics Virtually every university in the United States now has an Institutional Review Board that must pre-approve studies using human subjects High-impact social-science studies that resort to deception and cause stress are subject to scrutiny and may not be approved 41 The Belmont Report Issued by the U.S. National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research, 1978 Basic ethical principles and guidelines for the protection of human subjects of research: – Respect for persons – Consent – Beneficence – Justice 42 Respect for Persons Choices of autonomous individuals should be respected People incapable of making their own choices should be protected Voluntary subjects need adequate information for decision- making Consent Individuals should consent to participate in studies and those who cannot give their consent, such as children, people with diminished abilities, and prisoners, need to be protected. 43 Beneficence Participation in research is associated with a favorable balance of potential benefits and harms Maximize possible benefits, minimize potential harm Justice Ensuring reasonable, non-exploitative, and well-considered procedures are administered fairly — the fair distribution of costs and benefits to potential research participants — and equally. 44 Types of Risks and Benefits Risks or harms and benefits may be physical (pain or injury), psychological, social, economic, or legal. – Risk to a subject’s personal social standing, privacy, personal values and beliefs, links to family & wider community and work – Potential for harm can also arise from revealing information related to illegal, sexual, political or deviant behavior Risks or benefits of research may apply to individual participants, families, groups or organizations, communities, or nations.  not just to participant Risks and benefits to the research participant usually carry the most weight. 45 Point of Contention: Deception Passive deception (or omission) is the withholding or omitting of information; the researcher intentionally does not tell participants some information about the study. Active deception (or commission) is the presenting of misinformation about the study to participants. The most common form of active deception is misleading participants about the specific purpose of the study. 46 Justified Deception Deception may be justified in terms of some significant benefit that outweighs the risk to the participants. The researcher must consider all alternatives to deception and must justify the rejection of any alternative procedures. Question: Would lurking as a member in web forums or Facebook groups be acceptable to collect data? 47 Striking a Balance Advancing science and protecting the individuals with whom researchers work are two of experimental researcher’s core values Since every interaction with another human being raises at least the possibility of harm, avoiding all harm entirely would end all research Allowing research to move forward with no restrictions whatsoever would risk harms that researchers agree could not be justified by the knowledge gained For this reason, we must find a point on a continuum that strikes the correct balance 48 Guidelines for (Your) Research Work Rigor – Act with skill and care; keep skills up to date – Prevent corrupt practice and declare conflicts of interest – Respect and acknowledge the work of other scientists Respect – Ensure that research is justified and lawful – Minimize impact on people, animals and the environment Responsibility – Discuss issues science raises for society – Do not mislead; present evidence honestly 49 Industrial Research Ethics 50 Facebook’s Social Contagion Experiment What was that all about? https://www.theguardian.com/technology/2014/jul/02/facebook-apologises-psychological-experiments-on-users 51 Facebook’s Emotional Contagion Experiment – Setup Experiment manipulated the extent to which Facebook users (N = 689,003) were exposed to emotional expressions in their News Feed Manipulation: Researchers hid "a small percentage" of emotional words from peoples’ news feeds, without their knowledge, to test what effect that had on the statuses or "likes" that they then posted or reacted to Two dependent variables were examined pertaining to emotionality expressed in people’s own status updates: the percentage of all words produced by a given person that were either positive or negative during the experimental period. – In total, over 3 million posts were analyzed, containing over 122 million words, 4 million of which were positive (3.6%) and 1.8 million negative (1.6%). – Experiments took place for 1 week (January 11–18, 2012); randomly selected users 52 Facebook’s Emotional Contagion Experiment – Findings When positive expressions were reduced, users produced fewer positive posts and more negative posts; when negative expressions were reduced, the opposite pattern occurred.  These results indicate that emotions expressed by others on Facebook influence our own emotions, constituting experimental evidence for massive-scale contagion via social networks.  Contagion does not need personal (offline) contact 53 Facebook’s Version of Informed Consent – As Included in the Paper Linguistic Inquiry and Word Count software was adapted to run on the Hadoop Map/Reduce system and in the News Feed filtering system, such that no text was seen by the researchers As such, it was consistent with Facebook’s Data Use Policy, to which all users agree prior to creating an account on Facebook, constituting informed consent for this research. 54  Is the response by PNAS sufficient? E.g., no comment on whether data use policy is actually read and understood by users Note: The “Common Rule” is a set of regulations that essentially enshrine the recommendations of the Belmont report [https://www.pnas.org/content/111/29/10779.1] 55 Facebook’s Data Use Policy I. Product research and development: We use the information we have to develop, test and improve our Products, including by conducting surveys and research, and testing and troubleshooting new products and features. [https://www.facebook.com/policy.php; July 2019] 56 Facebook’s Data Use Policy II. Research and innovate for social good. We use the information we have (including from research partners we collaborate with) to conduct and support research and innovation on topics of general social welfare, technological advancement, public interest, health and well-being. For example, we analyze information we have about migration patterns during crises to aid relief efforts. Learn more about our research programs. 57 Facebook’s Data Use Policy III. Researchers and academics. We also provide information and content to research partners and academics to conduct research that advances scholarship and innovation that support our business or mission, and enhances discovery and innovation on topics of general social welfare, technological advancement, public interest, health and well-being. 58 However, what was the policy in 2012? “for internal operations, including … research” was only added to its data use policy in May 2012, four months after Facebook performed the published experiments In January 2012, the policy was different: … as part of our efforts to keep Facebook safe and secure; to provide you with location features and services, like telling you and your friends when something is going on nearby; to measure or understand the effectiveness of ads you and others see; to make suggestions to you and other users on Facebook, such as: suggesting that your friend use our contact importer because you found friends using it, suggesting that another user add you as a friend because the user imported the same email address as you did, or suggesting that your friend tag you in a picture they have uploaded with you in it. [January 2012] https://www.theguardian.com/technology/2014/jul/01/facebook-data-policy-research-emotion-study 59 Should Companies Do Such Research and Under What Rules? Massive amounts of data and money Boatloads of PhDs 60 Takeaways Research should be conducted to study and make progress on pressing issues in medicine, informatics and social science etc. – Should be conducted ethically – Difficult history – Lots of rules, and intuitions (IRBs) have evolved A „lot of action“ now focused on Internet platforms and services – Self-regulated – Poor notice and consent practices – Choice about participating in experiment not possible. Leaving platform? Lessons learned for your own work 61 The final lecture. Later today wrap-up and discussion. 62