HS2024 Lecture 4 - Knowing Humans, Selves PDF

Artificial Intelligence & Human Values Fall 2024, ETH Zürich Prof. Margarita Boenig-Liptsin Lecture 4: Knowing Humans, Knowing Selves 2024 Nobel Prize in Physics to… John Hopfield and Geoffrey Hinton The Nobel Prize,...

Artificial Intelligence & Human Values Fall 2024, ETH Zürich Prof. Margarita Boenig-Liptsin Lecture 4: Knowing Humans, Knowing Selves 2024 Nobel Prize in Physics to… John Hopfield and Geoffrey Hinton The Nobel Prize, October 8, 2024 NYT, May 1, 2023 Science, May 20, 2024 From last week AI in learning and knowledge-making brings into focus 1. Practices of how we think (and learn) 2. Student-teacher relationships 3. Dominant ways of knowing From last week Microsoft Research, 2009. From last week "Back" to the age of Computation? Prof. Torsten Hoefler at the Inauguration of "Alps" Research Symposium, September 13, 2024 From last week "The End of Theory" "The more we learn about biology, the further we find ourselves from a model that can explain it. [...] We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science [i.e. models/theories] cannot" Anderson, Wired, 2008 From last week A “new empiricism” Full resolution, granularity without sampling (“n=all”) Induction over deduction Beyond hypothesis testing Instrumentalist - Prediction and correlation over (causal) explanation and understanding “Data speak for themselves”; ML/AI find new patterns undetectable to human sight From last week Critique of "end of theory" data always selective data always theory-laden (classification) Why was the data collected? With what instruments? How was it labelled and organized? The meaning of data, models, and inferences are always situated, always involve interpretation and domain knowledge From last week Big Data, AI and conditions of knowledge at-large “Big Data” as sociotechnical fact (high volume,variety, velocity) as imaginary and “world schema” (“dataism”) Context or situation of data: world as laboratory How is data valued in different contexts? Political economy of data Balance shifts from academic to corporate context The Economist, February 2010 From last week "insist on a better account of the world," p. 579 What's needed: "knowledge potent for constructing worlds less organized by axes of domination", p. 585 Haraway, Situated Knowledges, 1988 From last week Situated AI? Domain-specific AI "We also don't want to theorize the world, much Reproducibility less act within it, in terms of Global Systems, but Transparency we do need an earth-wide network of connections, including the ability partially to translate knowledges among very different–and power-differentiated–communities" – Haraway, Situated Knowledges, p.580 Nature, September 14, 2023 From last week 3. Dominant ways of knowing What ways of knowing does AI depend upon? How are these integrated within a given discipline? What are the affordances and limits of ways of knowing with AI (for that discipline)? How can these be made visible and counter-balanced?  Questions for today's lecture How is AI involved in building knowledge about a person? About oneself? What is the history of how people have come to depend upon digital tools for self-knowledge and identity-making? What does it mean to consider “the human” in the process of designing AI systems? "Your Algorithm Doesn't Know Me" The Guardian, August 17, 2020 “That data is my artwork, that’s my life. It feels like my identity” - Ms. Ortiz, artist, UChicago News Stability.ai, August 22, 2022  “We’re taking our consent back,” Ms. Ortiz said. A.I.-generating tools, many of which charge users a fee to generate images, “have data that doesn’t belong to Shan them,”etsheal.,said. Feb 8, 2023, “That Arxiv data Mitchum, Feb 13, 2023, UChicago News is my artwork, that’s my life. It feels like my identity.”  Property Frame Context of Snowden Class Action Complaint, revelations (c. 2013) and San Francisco Court, Jan 13, 2023 GDPR (2018)  “That data is my artwork, that’s my life. It feels like my identity” - Ms. Ortiz, artist, "Your UChicago News Algorithm Doesn't Know Me" From allocative to representational harms  Identity Life-shaping and socially conditioned aspects of selfhood, such as gender, race, class, education, handicap status, income, immigration status. Informed and comprised of one's life experience, culture, ethics, style… Identity is not only about how you see yourself but also how (others in) society see(s) and treat(s) you (positionality). Identity configures and is configured with technology.  GDPR "Personal data" “any information relating to an identified or identifiable natural person” https://gdpr.eu/eu-gdpr-personal-data/ Not just human data, personal data Deriving from individuals; having its origins in people's bodies and behaviors. - Biometrics (body measurements and calculations), both physiological (e.g. facial measurements, DNA, irises, fingerprints) and behavioral (e.g. voice, gait, typing rhythm) - Profiles, demographic data from bank accounts, medical records, employment data. - Web searches and sites visited, likes, purchase histories - Tweets, texts, emails, phone calls, photos, videos, coordinates of real-world locations - Intellectual products such as artwork or writing? Volunteered (e.g. social network profile) Observed (e.g. location data) Inferred (e.g. credit score) "Accumulated through the living of ordinary life" Image: (Rebecca Lemov, Osiris, "Data Histories, 2017, p.23). Aadhar registration Becoming an "informational person" (Colin Koopman, How We Became Our Data, 2019)  "In a 2016 study, a man who had used a brain stimulator to treat his depression for seven years reported in a focus group that he began to wonder whether the way he was interacting with others — for example, saying something that, in retrospect, he thought was inappropriate — was due to the device, his depression or whether it reflected something deeper about himself. He said: “It blurs to the point where I'm not sure... frankly, who I am.” Yuste et al., Nature Comment, November 8, 2017 NYT, April 17, 2024 European Parilament, MIT Technology Review, October 4, 2024 July 24, 2024  Personal data - Derived from persons - Informing persons Genetic sequence in digital form. Who owns genetic information -- First portrait photograph, Robert Cornelius, the person from whose body it is 1839. Who owns the photograph -- the taken or the scientist/organization photographer or the sitter? that sequences and digitizes it? What's at stake in personal data? “[W]hen people tell us, ‘No, this is not where you are from,’ and your own blood says so — it is confusing to us,” Rex Tilousi said. “It hurts the elders who have been telling these stories to our grandchildren.” [...] "Our coming from the canyon, that is the basis of our sovereign rights,” said Edmond Tilousi, the tribe’s vice chairman." (A. Harmon, April 21, 2010, NYT) Havasupai reclaiming blood samples. Images: Jim Wilson, New York Times  "[A]fter all who are Men? Is every featherless biped to be counted a man and brother?" - W.E.B. Du Bois (1899) Sanitation workers' strike, Beale Street, Memphis, Tennessee, March 29, 1968. Bettmann Collection / Getty Images.  "Data brokers" Appears in FastCompany, March 2, 2019. Image: Cracked Labs Pre-digital histories of personal information (Drawing upon work of Rebecca Lemov, including “Big Data is People!” Aeon) “Mining subjectivities” 20th century, social scientists (sociologists, anthropologists, geographers) interested in increasingly more intimate details about human beings and their behaviors document and create data banks to “store" dreams, hallucinations, idle memories Image: Lemov, "What Escapes the Total Archive," Limn Projects animated by “fantasy of total information”: collecting the unremarkable, mundane, ordinary with the idea that something interesting and worthwhile could emerge desire to make the subjective more objective and more actionable (in order to optimize or automate behaviors — helping people become “better,” “healthier,” “efficient”) Lessons from historical projects The tendency to collect “human materials” leads to seeing humans as materials — reducing people to the information that they provide Long history of becoming "informational persons"  Quantified Self and conversing with chatbots "I trained an AI chatbot on my childhood journal entries, so that I could engage in real-time dialogue with my "inner child". I kept diaries for about 10+ years of my life, writing almost everyday — about my dreams, fears, and secrets. The content ranged from complaining about homework, to giddiness I felt from talking to my crush. Some days were very mundane, some rather insightful. In any case, there Young Michelle AI Chatbot, was a lot of it: fantastic, ripe data source for my Michelle Huang experiment." Quantifiedself.com "Self Image: Till Lauer; The — Michelle Huang Knowledge Through Numbers" Economist, 2022 In a world where personal data has value for corporations, governments, and individuals, what practices can help to respect its complex value? What does it mean to consider “the human” in the process of designing AI systems?  “I think content like the video we made is good for humanity,” Ardayfio said. “Because of this video, millions of people have seen the potential of reverse-image search technology and LLMs in a relatively wholesome, harmless way. But more importantly, they now know how to take control of their own data against people who aren’t like us and have bad interests! Bad actors have always been able to do reverse image searches, use LLMs, scrape websites, etc., but we’re exposing what these people can do and how to protect yourself.” – Caine Ardayfio, in Gizomodo AnhPhu Nguyen on X, September 30, 2024  Key Takeaways (I) Personal data are data that "are any information relating to an identified or identifiable natural person" (GDPR definition) Personal data matter because they 1) are derived from individual bodies and behaviors 2) inform personhood What counts as "personal data" is context dependent (where the context can be historical, cultural, legal, technical, situational).  Key Takeaways (II) Changing personhood: Consolidation of "informational personhood"; Projects create a constant state of experimentation where participants are research subjects Self-understanding (e.g relation between one's genetic info and one's person); citizenship; relationship to corporations Concern about being treated as mere data points, being subject to influence by those who control the personal data Rights: If people are at the origin of the data, what rights do they have to it? Who decides: What is private? What is shared? What is archived? What is forgotten? What is held in abeyance for the future? Transcending the contexts in which data originated in bodies or behaviors  Coming Up AIxKnowledges Unit Turing, Alan. 1950. “Computing Machinery and Intelligence.” Mind 59(236) Dick, Philip K. 1986. Do Androids Dream of Electric Sheep? [excerpt]  Extra! Should you wish to know… De-identification / Re-identification Processes used to prevent a person to be linked to information, e.g. deleting or masking name or social security numbers, suppressing date of birth or zip code De-identification is not foolproof — information can be re-identified e.g., by combining anonymous / de-identified data sets with publicly available information (zip code, date of birth, gender). 87% of Americans uniquely identified by zip code, date of birth, gender (Sweeney). Latanya Sweeney (2000)