Artificial Intelligence and Diversity for All

Software is an infrastructure of all industries and societies around the world, serving global users despite social differences, including race, gender, class, ethnicity, sexuality, and nationality.

In the era of Artificial Intelligence (AI) and increasing automation in software industry, the role of humans is even more emphasized across age, culture, and gender. However, the engagement of people in software and AI engineering is not uniform. It is important to address the diversity gap in software engineering urgently when new AI intensive software systems are being created because there is a risk that AI generated software perpetuates sexist and racist assumptions and ideologies.

The concept of intersectionality explores the interconnectedness of social differences, including race, gender, class, ethnicity, sexuality, and nationality. The goal of this lecture is to discuss the state of the art about diversity issues in core topics of AI and software engineering.

See also https://speakers.acm.org/speakers/jaccheri_10303

1. AI and Diversity for All Letizia Jaccheri

2. The Distinguished Speakers Program is made possible by For additional information, please visit http://dsp.acm.org/

3. About ACM ACM, the Association for Computing Machinery ( www.acm.org ), is the premier global community of computing professionals and students with nearly 100,000 members in more than 170 countries interacting with more than 2 million computing professionals worldwide. OUR MISSION:  We help computing professionals to be their best and most creative. We connect them to their peers, to what the latest developments, and inspire them to advance the profession and make a positive impact on society. OUR VISION:  We see a world where computing helps solve tomorrow’s problems – where we use our knowledge and skills to advance the computing profession and make a positive social impact throughout the world. I am proud to be an ACM Member.

4. Session 1 – AI – inspiration books and videos Session 2 – AI - history and terminology Session 3 – Software Engineering and diversity Session 4 – AI for or against all?

5. Kimberlé Crenshaw Bias

6. Literature Female roles in literature Nora (Ibsen) Literature for women women's magazines, Disney (Pochahontas, The Little Mermaid) Women who create literature Jane Austen, Sigrid Undset, Elena Ferrante

7. AI Female roles in AI Siri voice Avatar AI for/ against women For – Menstruation Apps, Designing Software to Prevent Child Marriage Globally , Tappetina Against - automatic processing of CVs Women creating AI Fei Fei Li Francesca Rossi

8. Questions Which of these books do you relate to? Which ones make you want to read/listen? Letiziajaccheri.org

9. Session 1 – AI – inspiration books and videos Session 2 – AI - history and terminology Session 3 – Software Engineering and diversity Session 4 – AI for or against all?

10. Definitions AI is a field of study (and research field) within computer science that develops and studies intelligent machines AI stands for a computer system that performs tasks that typically require human intelligence, such as recognizing speech, making decisions and identifying patterns Generatively create new content (sound, code, images, text, video) Machine learning is part of AI

11. History Artificial intelligence was founded as an academic discipline in 1956 1950 - 60 first AI programs 1970 expert systems 1980 neural networks 1990 autonomous robots 2,000 self-driving cars 2010 AI-powered assistants 2020 Advanced AI in healthcare, finance, transport, art

12. Why now ? Hardware – software – data GPT – Generativ Pretrained Transformer autumn 2022

13. hardware From Kilobyte til Petabyte 10 15

14. software

15. CHAT GPT 4 has been trained on almost all text ever written data 10 13

16. IT system Humans Develop , test, use AI system Generative AI Use tools

17. - GPT-3 has an estimated training time of 355-GPU-years and an estimated training cost of $4.6 million . - If we trained GPT-3 at IDUN, it would take 355/36 = 10 years

18. Discussion questions What new words have you learned? What are your questions about AI?

19. Session 1 – AI – inspiration books and videos Session 2 – AI - history and terminology Session 3 – Software Engineering and diversity Session 4 – AI for or against all?

20. Software Engineering Gender Analysis and Design | Empirical software engineering | Software quality | Architecture | Processes | AI and SE | Human factors in SE Gender and sex | Non-binary | LGBT+ rights | #metoo 2017 | Same-sex marriage 2001 | Intersectionality – triply | feminism

21. Kimberlé Crenshaw Bias ( bug , error )

22. Amazon created a recruitment tool that proved to be discriminating against women specifically J. Dastin , “Amazon scraps secret AI recruiting tool that showed bias against women,” in Ethics of data and analytics , Auerbach Publications, 2022, pp. 296–299.

23. Facebook’s job advertisement algorithm reached out to specific users based on their race, gender, and religion . Moreover , women were presented with stereotypical feminine jobs, such as secretaries or nurses. Such algorithms enhance sexism and racist attitudes in the labor environment . M. Ali, P. Sapiezynski , M. Bogen , A. Korolova , A. Mislove , and A. Rieke , “ Discrimination through optimization : How facebook’s ad delivery can lead to biased outcomes ,” Proceedings of the ACM on human-computer interaction, vol. 3, no. CSCW, pp. 1–30, 2019.

24. 20% 29% Female ICT students in 2021

25. I nformation M entoring N etwork A nti bias training Interventions

26. Norwegian and European best Practices ADA IDUN EUGAIN Horizon CRAFT Erasmus + Women Stem Up ACM WomENcourage Abelia Tech Kvinner

27. Burnett, M., Stumpf , S., Macbeth, J., Makri , S., Beckwith , L., Kwan, I., Peters, A. and Jernigan , W., 2016. GenderMag : A method for evaluating software's gender inclusiveness .  Interacting with computers ,  28 (6), pp.760-787.

28. G. Catolino , F. Palomba, D. A. Tamburri, A. Serebrenik and F. Ferrucci, "Gender Diversity and Women in Software Teams: How Do They Affect Community Smells ?,"  2019 IEEE/ACM 41st International Conference on Software Engineering: Software Engineering in Society (ICSE-SEIS) , Montreal, QC, Canada, 2019, pp. 11-20

29. RQ How do biases in the workforce impact biases in software? IT system workforce Develop , test, use AI system Generative AI Use tools

30. RQ How do biases in the workforce impact biases in software? Y. Wang and D. Redmiles , “ Implicit gender biases in professional software development : An empirical study ,” in 2019 IEEE/ACM 41st International Conference on Software Engineering: Software Engineering in Society (ICSE-SEIS), 2019, pp . 1–10.

31. Implicit Association Test (IAT) The Implicit Association Test (IAT) is a psychological tool used to measure the strength of automatic associations between mental concepts, such as between certain social groups (e.g., race, gender, age) and evaluations (e.g., good or bad) or stereotypes (e.g., athletic, smart). The test is designed to uncover implicit biases—attitudes or stereotypes that people may be unaware of or unwilling to disclose.

32. Modern Sexism Scale (MSS) measure beliefs as The belief that gender discrimination is no longer a significant issue. Opposition to policies designed to reduce gender inequality, like affirmative action. Resistance to feminist movements or gender equality initiatives, often under the guise of advocating for fairness or merit-based systems.

33. RQ How do biases in the workforce impact biases in software ? A. Hannak , G. Soeller , D. Lazer, A. Mislove , and C. Wilson, “Measuring price discrimination and steering on e-commerce web sites,” in Proceedings of the 2014 conference on internet measurement conference, 2014, pp. 305–318.

34. Session 1 – AI – inspiration books and videos Session 2 – AI - history and terminology Session 3 – Software Engineering and diversity Session 4 – AI for or against all?

35. openart.ai/ create

36. 2012: 17% 2023: 19,4%

37. Hvorfor ? IT system Mennesker lager, tester, bruker KI system Generativ-KI bruke verktøyene 50% 5%

38. Threats False statements, false faces, false messages. There will be more of all this. Old systems, old stereotypes are magnified - if we don't take action Automatic processing of CVs The training data

39. AI for all We cannot change old networks , we can make new ones around AI AI for women https://irthapp.com/

40. 1.8.2024

41. Discussion questions What can I do? What do I want to do?

42. TDT4290 Customer Driven Project https://tinyurl.com/2x5y5mnk Customer defines the project – The teaching team, the students learn together with the customer 2023 Artificial Intelligence Sustainability Gender and Diversity Thale Kuvås Solberg (Q-Free) ACM womENcourage 2023