ethics fairness in ai ethics fairness in ai enabled
play

ETHICS & FAIRNESS IN AI- ETHICS & FAIRNESS IN AI- ENABLED - PowerPoint PPT Presentation

Apr 13, 2023 •642 likes •1.41k views

ETHICS & FAIRNESS IN AI- ETHICS & FAIRNESS IN AI- ENABLED SYSTEMS ENABLED SYSTEMS Christian Kaestner (with slides from Eunsuk Kang) Required reading: R. Caplan, J. Donovan, L. Hanson, J. Matthews. " Algorithmic Accountability:

  1. ETHICS & FAIRNESS IN AI- ETHICS & FAIRNESS IN AI- ENABLED SYSTEMS ENABLED SYSTEMS Christian Kaestner (with slides from Eunsuk Kang) Required reading: ฀ R. Caplan, J. Donovan, L. Hanson, J. Matthews. " Algorithmic Accountability: A Primer ", Data & Society (2018). 1

  2. LEARNING GOALS LEARNING GOALS Review the importance of ethical considerations in designing AI-enabled systems Recall basic strategies to reason about ethical challenges Diagnose potential ethical issues in a given system Understand the types of harm that can be caused by ML Understand the sources of bias in ML Analyze a system for harmful feedback loops 2

  3. OVERVIEW OVERVIEW Many interrelated issues: Ethics Fairness Justice Discrimination Safety Privacy Security Transparency Accountability Each is a deep and nuanced research topic. We focus on survey of some key issues. 3

  4. ETHICAL VS LEGAL ETHICAL VS LEGAL 4 . 1

  5. In September 2015, Shkreli received widespread criticism when Turing obtained the manufacturing license for the antiparasitic drug Daraprim and raised its price by a factor of 56 (from USD 13.5 to 750 per pill), leading him to be referred to by the media as "the most hated man in America" and "Pharma Bro". -- Wikipedia " I could have raised it higher and made more profits for our shareholders. Which is my primary duty. " -- Martin Shkreli 4 . 2

  6. Speaker notes Image source: https://en.wikipedia.org/wiki/Martin_Shkreli#/media/File:Martin_Shkreli_2016.jpg

  7. TERMINOLOGY TERMINOLOGY Legal = in accordance to societal laws systematic body of rules governing society; set through government punishment for violation Ethical = following moral principles of tradition, group, or individual branch of philosophy, science of a standard human conduct professional ethics = rules codified by professional organization no legal binding, no enforcement beyond "shame" high ethical standards may yield long term benefits through image and staff loyalty 4 . 3

  8. WITH A FEW LINES OF CODE... WITH A FEW LINES OF CODE...

  9. 4 . 4

  10. THE IMPLICATIONS OF OUR CHOICES THE IMPLICATIONS OF OUR CHOICES “Update Jun 17: Wow—in just 48 hours in the U.S., you recorded 5.1 years worth of music—40 million songs—using our doodle guitar. And those songs were played back 870,000 times!“ 4 . 5

  11. Amazing version of Hey Jude played through the Google Doodle 'Les Paul' guitar - June 9 Amazing version of Hey Jude played through the Google Doodle 'Les Paul' guitar - June 9 … 4 . 6

  12. CONCERNS ABOUT AN AI CONCERNS ABOUT AN AI FUTURE FUTURE 5 . 1

  13. SAFETY SAFETY 5 . 2

  14. SAFETY SAFETY Tweet 5 . 3

  15. SAFETY SAFETY Tweet 5 . 4

  16. ADDICTION ADDICTION 5 . 5

  17. Speaker notes Infinite scroll in applications removes the natural breaking point at pagination where one might reflect and stop use.

  18. ADDICTION ADDICTION

  19. 5 . 6

  21. 5 . 7

  22. MENTAL HEALTH MENTAL HEALTH

  23. 5 . 8

  24. 5 . 9

  25. SOCIETY: UNEMPLOYMENT ENGINEERING / SOCIETY: UNEMPLOYMENT ENGINEERING / DESKILLING DESKILLING 5 . 10

  26. Speaker notes The dangers and risks of automating jobs. Discuss issues around automated truck driving and the role of jobs. See for example: Andrew Yang. The War on Normal People. 2019

  27. SOCIETY: POLARIZATION SOCIETY: POLARIZATION 5 . 11

  28. Speaker notes Recommendations for further readings: https://www.nytimes.com/column/kara-swisher , https://podcasts.apple.com/us/podcast/recode-decode/id1011668648 Also isolation, Cambridge Analytica, collaboration with ICE, ...

  29. WEAPONS, SURVEILLANCE, SUPPRESSION WEAPONS, SURVEILLANCE, SUPPRESSION

  30. 5 . 12

  31. DISCRIMINATION DISCRIMINATION Tweet 5 . 13

  32. DISCRIMINATION DISCRIMINATION 5 . 14

  33. DISCRIMINATION DISCRIMINATION Unequal treatment in hiring, college admissions, credit rating, insurance, policing, sentencing, advertisement, ... Unequal outcomes in healthcare, accident prevention, ... Reinforcing patterns in predictive policing with feedback loops Technological redlining 5 . 15

  34. ANY OWN EXPERIENCES? ANY OWN EXPERIENCES? 5 . 16

  35. SUMMARY -- SO FAR SUMMARY -- SO FAR Safety issues Addiction and mental health Societal consequences: unemployment, polarization, monopolies Weapons, surveillance, suppression Discrimination, social equity Many issues are ethically problematic, but some are legal. Consequences? Intentional? Negligence? Unforeseeable? 5 . 17

  36. FAIRNESS FAIRNESS 6 . 1

  37. LEGALLY PROTECTED CLASSES (US) LEGALLY PROTECTED CLASSES (US) Race (Civil Rights Act of 1964) Color (Civil Rights Act of 1964) Sex (Equal Pay Act of 1963; Civil Rights Act of 1964) Religion (Civil Rights Act of 1964) National origin (Civil Rights Act of 1964) Citizenship (Immigration Reform and Control Act) Age (Age Discrimination in Employment Act of 1967) Pregnancy (Pregnancy Discrimination Act) Familial status (Civil Rights Act of 1968) Disability status (Rehabilitation Act of 1973; Americans with Disabilities Act of 1990) Veteran status (Vietnam Era Veterans' Readjustment Assistance Act of 1974; Uniformed Services Employment and Reemployment Rights Act) Genetic information (Genetic Information Nondiscrimination Act) Barocas, Solon and Moritz Hardt. " Fairness in machine learning ." NIPS Tutorial 1 (2017). 6 . 2

  38. REGULATED DOMAINS (US) REGULATED DOMAINS (US) Credit (Equal Credit Opportunity Act) Education (Civil Rights Act of 1964; Education Amendments of 1972) Employment (Civil Rights Act of 1964) Housing (Fair Housing Act) ‘Public Accommodation’ (Civil Rights Act of 1964) Extends to marketing and advertising; not limited to final decision Barocas, Solon and Moritz Hardt. " Fairness in machine learning ." NIPS Tutorial 1 (2017). 6 . 3

  40. 6 . 4

  41. HARMS OF ALLOCATION HARMS OF ALLOCATION Withhold opportunities or resources Poor quality of service, degraded user experience for certain groups Other examples?

  42. Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification , Buolamwini & Gebru, ACM FAT* (2018). 6 . 5

  43. HARMS OF REPRESENTATION HARMS OF REPRESENTATION Reinforce stereotypes, subordination along the lines of identity Other examples? Latanya Sweeney. Discrimination in Online Ad Delivery , SSRN (2013).

  44. 6 . 6

  45. IDENTIFYING HARMS IDENTIFYING HARMS Multiple types of harms can be caused by a product! Think about your system objectives & identify potential harms. Swati Gupta, Henriette Cramer, Kenneth Holstein, Jennifer Wortman Vaughan, Hal Daumé III, Miroslav Dudík, Hanna Wallach, Sravana Reddy, Jean GarciaGathright. Challenges of incorporating algorithmic fairness into practice , FAT* Tutorial, 2019. ( slides ) 6 . 7

  46. THE ROLE OF REQUIREMENTS ENGINEERING THE ROLE OF REQUIREMENTS ENGINEERING Identify system goals Identify legal constraints Identify stakeholders and fairness concerns Analyze risks with regard to discrimination and fairness Analyze possible feedback loops (world vs machine) Negotiate tradeoffs with stakeholders Set requirements/constraints for data and model Plan mitigations in the system (beyond the model) Design incident response plan Set expectations for offline and online assurance and monitoring 6 . 8

  47. WHY CARE ABOUT FAIRNESS? WHY CARE ABOUT FAIRNESS? Obey the law Better product, serving wider audiences Competition Responsibility PR Examples? Which argument appeals to which stakeholders? Swati Gupta, Henriette Cramer, Kenneth Holstein, Jennifer Wortman Vaughan, Hal Daumé III, Miroslav Dudík, Hanna Wallach, Sravana Reddy, Jean GarciaGathright. Challenges of incorporating algorithmic fairness into practice , FAT* Tutorial, 2019. ( slides ) 6 . 9

  48. CASE STUDY: COLLEGE ADMISSION CASE STUDY: COLLEGE ADMISSION Objective: Decide "Is this student likely to succeed"? Possible harms: Allocation of resources? Quality of service? Stereotyping? Denigration? Over-/Under-representation? 6 . 10

  49. NOT ALL DISCRIMINATION IS HARMFUL NOT ALL DISCRIMINATION IS HARMFUL Loan lending: Gender discrimination is illegal. Medical diagnosis: Gender-specific diagnosis may be desirable. Discrimination is a domain-specific concept! Other examples? 6 . 11

  50. ON TERMINOLOGY ON TERMINOLOGY Bias and discrimination are technical terms in machine learning selection bias reporting bias bias of an estimator inductive/learning , , , bias discrimination refers to distinguishing outcomes (classification) The problem is unjustified differentiation, ethical issues practical irrelevance moral irrelevance 6 . 12

  51. SOURCES OF BIAS SOURCES OF BIAS 7 . 1

  52. WHERE DOES THE BIAS COME FROM? WHERE DOES THE BIAS COME FROM? Caliskan et al., Semantics derived automatically from language corpora contain human-like biases , Science (2017). 7 . 2

  53. SOURCES OF BIAS SOURCES OF BIAS Tainted examples / historical bias Skewed sample Limited features Sample size disparity Proxies Barocas, Solon, and Andrew D. Selbst. " Big data's disparate impact ." Calif. L. Rev. 104 (2016): 671. Mehrabi, Ninareh, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. "] ( https://arxiv.org/pdf/1908.09635.pdf." arXiv preprint arXiv:1908.09635 (2019). 7 . 3

Recommend

INTRO TO ETHICS AND INTRO TO ETHICS AND FAIRNESS FAIRNESS Eunsuk Kang Required reading: R.

INTRO TO ETHICS AND INTRO TO ETHICS AND FAIRNESS FAIRNESS Eunsuk Kang Required reading: R.

INTRO TO ETHICS AND INTRO TO ETHICS AND FAIRNESS FAIRNESS Eunsuk Kang Required reading: R. Caplan, J. Donovan, L. Hanson, J. Matthews. "Algorithmic Accountability: A Primer", Data & Society (2018). 1 LEARNING GOALS LEARNING

732 views • 48 slides
Fairness, Ethics, and Machine Learning Prof. Mike Hughes Many ideas/slides attributable to:

Fairness, Ethics, and Machine Learning Prof. Mike Hughes Many ideas/slides attributable to:

Tufts COMP 135: Introduction to Machine Learning https://www.cs.tufts.edu/comp/135/2019s/ Fairness, Ethics, and Machine Learning Prof. Mike Hughes Many ideas/slides attributable to: Alexandra Chouldechova Moritz Hardt 3 Fairness: Unit

536 views • 34 slides
ETHICS IN GOVERNMENT : Ethics Overview for the Palliative Care Interdisciplinary Advisory

ETHICS IN GOVERNMENT : Ethics Overview for the Palliative Care Interdisciplinary Advisory

ETHICS IN GOVERNMENT : Ethics Overview for the Palliative Care Interdisciplinary Advisory Council David A. Reisman Chief Ethics Officer Health and Human Services Commission What is Ethics? W 2 Where do Ethics Laws W Come From? Ethics

533 views • 34 slides
Introduction to me What is Ethics? Ethics is about how we meet the challenge of doing the

Introduction to me What is Ethics? Ethics is about how we meet the challenge of doing the

Introduction to me What is Ethics? Ethics is about how we meet the challenge of doing the right thing when that will cost more than we want to pay The Josephson Institute of Ethics What is Business Ethics? Rules that govern how businesses

794 views • 30 slides
What Is Active Ethics? Vigorous application of professional ethics, ethics, and simply doing

What Is Active Ethics? Vigorous application of professional ethics, ethics, and simply doing

6/4/2012 Active Ethics: How Doing the Right Thing Helps Everyone Presented by Ron Klein, J.D., CFE Vice President Risk Management Counsel What Is Active Ethics? Vigorous application of professional ethics, ethics, and simply doing the

74 views • 5 slides
Ethics and Research Integrity Department of Government London School of Economics and Political

Ethics and Research Integrity Department of Government London School of Economics and Political

Ethics Ethics at LSE Ethics and Research Integrity Department of Government London School of Economics and Political Science Ethics Ethics at LSE 1 Ethics 2 Ethics at LSE Ethics Ethics at LSE 1 Ethics 2 Ethics at LSE Ethics Ethics at LSE

1.15k views • 29 slides
Media Fairness, Diversity 1 Outline Fairness (case studies, basic definitions) Diversity

Media Fairness, Diversity 1 Outline Fairness (case studies, basic definitions) Diversity

Online Social Networks and Media Fairness, Diversity 1 Outline Fairness (case studies, basic definitions) Diversity An experiment on the diversity of Facebook 2 Fairness, Non-discrimination To discriminate is to treat someone

1.03k views • 99 slides
Normative Ethics: Utilitarianism, Deontology, and Virtue Ethics Normative Ethics Applied

Normative Ethics: Utilitarianism, Deontology, and Virtue Ethics Normative Ethics Applied

Normative Ethics: Utilitarianism, Deontology, and Virtue Ethics Normative Ethics Applied ethics is popular because it is easy to see the application Applied ethics is unpopular because it is difficult to make progress those who are

701 views • 49 slides
ETHICS IN THE MILITARY CONTRACT AWARD PROCESS Ethics Final Presentation Com 563 Ethics for

ETHICS IN THE MILITARY CONTRACT AWARD PROCESS Ethics Final Presentation Com 563 Ethics for

ETHICS IN THE MILITARY CONTRACT AWARD PROCESS Ethics Final Presentation Com 563 Ethics for Professionals Michelle K. Robinson WSU Summer 2015 1. Summary of the Ethics Dilemma In June 2015, a story broke regarding a high ranking military

753 views • 6 slides
ETHICS AND OPENNESS IN GOVERNMENT Election Commissioners Annual Meeting January 22, 2016 Notice:

ETHICS AND OPENNESS IN GOVERNMENT Election Commissioners Annual Meeting January 22, 2016 Notice:

MISSISSIPPI ETHICS COMMISSION Post Office Box 22746 660 North Street, Suite 100-C Jackson, Mississippi 39225-2746 Telephone: 601-359-1285 Facsimile: 601-359-1292 www.ethics.state.ms.us info@ethics.state.ms.us ETHICS AND OPENNESS IN

185 views • 16 slides
Ethics 4 Everyone! Ethics 4 Everyone! Trust, Quality, Service and Value Trust, Quality, Service

Ethics 4 Everyone! Ethics 4 Everyone! Trust, Quality, Service and Value Trust, Quality, Service

Ethics 4 Everyone! Ethics 4 Everyone! Trust, Quality, Service and Value Trust, Quality, Service and Value Despite the codes of ethics, the ethics programs and special departments, corporations dont make the ultimate decisions about ethics.

583 views • 18 slides
What is fairness? The parties have not shown us, and I have not been able to discover . . . .

What is fairness? The parties have not shown us, and I have not been able to discover . . . .

What is fairness? The parties have not shown us, and I have not been able to discover . . . . statements of principled, well-accepted rules of fairness that should govern districting. - Justice Anthony Kennedy, Vieth v Jubelirer (2004) Is

1.23k views • 79 slides
ETHICS - CAN and SHOULD are Two Different Things Our Time Together WHY we speak about

ETHICS - CAN and SHOULD are Two Different Things Our Time Together WHY we speak about

ETHICS - CAN and SHOULD are Two Different Things Our Time Together WHY we speak about ethics in the insurance industry Why focus on ETHICS and not Morality A brief history of Ethics To whom do we owe Ethical decisions?

883 views • 51 slides
HOW CAN FAIRNESS OPINIONS PROVIDE VALUE TO SHAREHOLDERS? 4th OIV International Business Valuation

HOW CAN FAIRNESS OPINIONS PROVIDE VALUE TO SHAREHOLDERS? 4th OIV International Business Valuation

HOW CAN FAIRNESS OPINIONS PROVIDE VALUE TO SHAREHOLDERS? 4th OIV International Business Valuation Conference 30 November 2015 Milan, Italy Jay E. Fishman, FASA Jfishman@finresearch.com 1 WHAT IS A FAIRNESS OPINION? A fairness opinion is

484 views • 36 slides
Fairness in Machine Learning: Part I Privacy & Fairness in Data Science CS848 Fall 2019

Fairness in Machine Learning: Part I Privacy & Fairness in Data Science CS848 Fall 2019

Fairness in Machine Learning: Part I Privacy & Fairness in Data Science CS848 Fall 2019 Outline High Level View Recap Supervised Learning: Binary Classification Survey of Approaches to Fairness in Supervised Learning Warmup:

593 views • 35 slides
AI Ethics for AI Practitioners A design framework for building towards algorithmic justice Willie

AI Ethics for AI Practitioners A design framework for building towards algorithmic justice Willie

AI Ethics for AI Practitioners A design framework for building towards algorithmic justice Willie Costello The hazards of algorithmic systems Bias Discrimination Racism Sexism Surveillance The ideals of algorithmic systems Fairness

575 views • 28 slides
On Combining State Space Reductions with Global Fairness Assumptions Shaojie Zhang 1 Jun Sun 2 Jun

On Combining State Space Reductions with Global Fairness Assumptions Shaojie Zhang 1 Jun Sun 2 Jun

Background & Motivation Model Checking with Global Fairness Symmetry Reduction & Global Fairness Partial Order Reduction & Global Fairness Summary On Combining State Space Reductions with Global Fairness Assumptions Shaojie Zhang 1

402 views • 25 slides
Primary Areas of Jurisdiction for the Ethics Commission Ethics in Government Law Public

Primary Areas of Jurisdiction for the Ethics Commission Ethics in Government Law Public

MISSISSIPPI ETHICS COMMISSION MASBO 46 TH Annual Conference February 15, 2016 Ethics and Openness in School Districts MISSISSIPPI ETHICS COMMISSION Eight members serve staggered terms. Members appointed by Governor, Lieutenant Governor,

508 views • 47 slides
Fairness opinions (FO) and Italian Valuation Standards (PIV) Mauro Bini Chair of OIV

Fairness opinions (FO) and Italian Valuation Standards (PIV) Mauro Bini Chair of OIV

Fairness opinions (FO) and Italian Valuation Standards (PIV) Mauro Bini Chair of OIV Management Board Agenda 1. A poll on Fairness Opinions 2. The elusive definition of Fairness Opinion and the main issues 3. An example and evidences

630 views • 32 slides
Fairness in ML 2: Equal opportunity and odds Privacy & Fairness in Data Science CS848 Fall

Fairness in ML 2: Equal opportunity and odds Privacy & Fairness in Data Science CS848 Fall

Fairness in ML 2: Equal opportunity and odds Privacy & Fairness in Data Science CS848 Fall 2019 Slides adapted from https://fairmlclass.github.io/4.html 2 Outline Recap: Disparity impact Issues with Disparate Impact

5.56k views • 35 slides
Data ethics Data ethics is the study and evaluation of Data ethics data problems related to

Data ethics Data ethics is the study and evaluation of Data ethics data problems related to

Data ethics Data ethics is the study and evaluation of Data ethics data problems related to data, algorithms, and generation, recording, curation, processing, dissemination, sharing, & use to formulate Studies & & support

72 views • 6 slides
Society Expanding context: Fairness A simple problem:

Society Expanding context: Fairness A simple problem:

Society Expanding context: Fairness A simple problem: classification Hiring Loan College admission Definitions of fairness Individuals with I treat you similar abilities differently

1.08k views • 71 slides
Land Ethics and Earth Ethics: Making the Connections Sustainability Ethics, the Earth Charter,

Land Ethics and Earth Ethics: Making the Connections Sustainability Ethics, the Earth Charter,

Land Ethics and Earth Ethics: Making the Connections Sustainability Ethics, the Earth Charter, and Aldo Leopold Earth Charter International & the University of Wisconsin-Oshkosh October 8, 2015 1933 1939 June 1947 July 1947 We abuse

219 views • 11 slides
Flow Rate Fairness: Dismantling a Religion Bob Briscoe presented by Brighten Godfrey 1 Fairness

Flow Rate Fairness: Dismantling a Religion Bob Briscoe presented by Brighten Godfrey 1 Fairness

Flow Rate Fairness: Dismantling a Religion Bob Briscoe presented by Brighten Godfrey 1 Fairness in networks Flow rate equality! 911 Easily circumvented Doesnt even optimize for any metric of interest hotornot.com Fig. 1:

484 views • 17 slides

More recommend