A Conversation on Critical Approaches to AI and Pedagogy

Ngozi Harrison
Department of Information Studies, UCLA
Aug 22, 2024

About Me

2nd year PhD Student in Department of Information Studies, UCLA
UCLA Cota-Robles Fellow and Center for Race and Digital Justice Fellow
Previously Creative Effectiveness Lead @ Google

Areas of Interest Critical AI Studies, Formal Reasoning and Computation, Race and Digital Justice, Knowledge Organization

Research Topic My research focuses on examining the mathematical, conceptual, and logical foundations of information systems and computation. I seek to explore non-western logics and modes of computation to develop liberatory frameworks for information science.

Overview

  • What is Artificial Intelligence
  • The Current AI Landscape
  • Critical Perspectives on Large Language Models
  • My Perspective on AI
  • AI as a Student Researcher
  • Pedagogy and Teaching in the Era of AI

On Methodology

No Technology is Neutral

We Have to Understand Technologies within Their Social Context and Avoid both Determinism and Dismissal.

Analyzing the Sociotechnical

We can understand any technology as an assemblage composed of the artifact, practices, and beliefs (Brock 2020). Analyzing technologies allows us to develop concepts that both have explanatory power and allow for a radical rethinking of the present toward more just futures. There many frameworks for the critical analysis of technology including Critical Race Theory, Feminist Analysis, Black Digital Feminism, Marxism and Political Economy, etc.

My goal in thinking about AI is to raise our collective critical sociotechnical consciousness where we don’t simply accept the terms set for us my corporate companies but refuse, reimagine, and reappropriate technologies so serve other purposes

From Algorithms to AI

According to Tufekci, algorithms are “computational agents who are not alive but act in the world” (Tufekci 2015). Benjamin extends this definition by calling them “formalizations” that enforce racialized meanings and contribute to the constitution of the world (Benjamin 2019).

What is AI?

A Working Definition of Artificial Intelligence

Artificial Intelligence as concept can be defined the ability of a computer system to perform tasks and that are traditionally thought to require human intelligence. Artificial Intelligence systems collections of data, algorithms, and various components that use the techniques of the discipline of AI and ML research.

What Are LLMs

Large Language models are a type of natural language processing model that are trained on massive data sets and use machine learning algorithms infer patterns and structures in human language. When LLMs generate responses they are not necessarily reasoning but using advanced probabilistic methods to respond in a way that mimic prior data in the dataset

Current Popular Models

ChatGPT-4
Latest Model from OpenAI
Initial Release in November 2022
Microsoft is a big investor and strategic partner

Claude 3.5
Developed by Anthropic, founders split from OpenAI
Initial Release in March 2023

Current Popular Models Cont'd

Gemini
Developed by Google
Initial Release in
Multimodal family of models, successor to Lambda and PaLM 2

LLaMA
Developed by Facebook/Meta
Open source
Released in 2023

Key Issues in AI Today

  • Ethics, Fairness, and Accountability in the use of AI systems
  • Misinformation/Disinformation
  • Environmental Impact of AI
  • Algorithmic bias
  • Intellectual Property and Regulation
  • AI, Military, and Surveillence


Estimates have shown training GPT-3 in Microsoft’s data centers can directly use 700,000 liters of clean freshwater (Li et al., 2023)

AI as we know it, is probably a bubble

How are you currently using AI?
What are the benefits?
What concerns do you have about AI?

Using LLMs as a Student Researcher

Increasingly there are new tools marketed toward students and researchers to make our lives easier. There are general tools such as ChatGPT and Claude in addition to more specific tools like Grammarly and search engines such as Perplexity.AI

How I Use AI

Due to some of the accuracy, and ethical/social concerns with AI systems I advocate for careful consideration and use of AI tools within research practice.
Current ways I use AI tools:

  • Finding research articles and sources
  • Editing and grammer
  • Brainstorming and outlining
  • Writing emails

Remember: there's more to AI tools than LLMs

AI and Ethics

“In essence, the future of AI and ethics should be concerned with rising global social and economic inequality, the repercussions that will emerge as an effect of climate change, and the ways in which AI will be used in the redistribution of global goods and services—from housing, to food, to border-crossing, and beyond. Broadly, those of us in the fields of digital social research must center the issues of social, political, and economic inequality as an orientation to studying lived experiences in relation to structures of power that algorithms, AI, and automated systems can overdetermine—rather than assuming that technology itself can be ethically perfected or that bias is a feature of a AI or externality that can be corrected or resolved.” (Le Bui and Noble, 2020, p. 10)

AI , Authorship & Copyright

Currently copyright law and regulation on AI is still unfolding, both globally and in the US. This means as scholars we have have to stay informed as things quickly change and we have some influence on what the future looks like.
Things to Know:

  • Generative Models and LLMs have been shown to be trained using copyrighted material
  • Nature, and Science are two example journals have judged writing our sources authored or coauthored using AI technologies do not meet their criteria
  • Elsevier requires disclosure of use of AI

Using AI in Qualitative Research Analysis

Davison et al. identify ethical issues in relation to the use of AI for analysis of qualitative data:

  • Data ownership and rights
  • Data privacy and transparency
  • Interpretive sufficiency
  • Biases manifested in Generative Artificial Intelligence (GAI)
  • Researcher responsibilities and agency
    The argue for building evolving living guidelines, researcher responsibility and transparency and robust ethical frameworks

How are you thinking about AI and Pedagogy?

Pedagogy and Teaching in the Era of AI

How Do We Prepare Students to Enter a World with AI

Students need information literacy, critical algorithm skills, and the tools to innovate using technologies and imagine just futures

This means not just teaching students how to use technology, but critically about AI technologies and imagine more just futures

AI technologies and systems are here for now, but what the future will look like is still being determined. We can imagine just and equitable futures

Sources

Brock, A. (2018). Critical technocultural discourse analysis. New Media & Society, 20(3), 1012–1030. https://doi.org/10.1177/1461444816677532
Davison, R. M., Chughtai, H., Nielsen, P., Marabelli, M., Iannacci, F., van Offenbeek, M., Tarafdar, M., Trenz, M., Techatassanasoontorn, A. A., Díaz Andrade, A., & Panteli, N. (n.d.). The ethics of using generative AI for qualitative data analysis. Information Systems Journal, n/a(n/a). https://doi.org/10.1111/isj.12504
Huber, L. P., Vélez, V. N., & Malagón, M. C. (2024). Charting methodological imaginaries: Critical Race Feminista Methodologies in educational research. International Journal of Qualitative Studies in Education, 37(5), 1263–1271. https://doi.org/10.1080/09518398.2024.2318296
Le Bui, M., & Noble, S. U. (2020). We’re Missing a Moral Framework of Justice in Artificial Intelligence: On the Limits, Failings, and Ethics of Fairness. In M. D. Dubber, F. Pasquale, & S. Das (Eds.), The Oxford Handbook of Ethics of AI (pp. 161–179). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780190067397.013.9
Li, P., Yang, J., Islam, M. A., & Ren, S. (2023). Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models (arXiv:2304.03271). arXiv. http://arxiv.org/abs/2304.03271
Tanksley, T. C. (2024). “We’re changing the system with this one”: Black students using critical race algorithmic literacies to subvert and survive AI-mediated racism in school. English Teaching: Practice & Critique, 23(1), 36–56. https://doi.org/10.1108/ETPC-08-2023-0102
Tufekci, Z. (2015). Algorithmic harms beyond Facebook and Google: Emergent challenges of computational agency. Colo. Tech. LJ, 13, 203.
Villa-Nicholas, M. (2019). Latinx Digital Memory: Identity Making in Real Time. Social Media + Society, 5(4), 2056305119862643. https://doi.org/10.1177/2056305119862643

Additional Reading and Links

Artificial Intelligence (Generative) Resources from Georgetown Library
Essential books by Scholars of Color scholars on tech, science, & race
Data & Society
Hugging Face
Center for Race and Digital Justice
A Sociotechnical Approach to AI Policy
Comments to Consumer Financial Protection Bureau regarding Collection and Sale of Consumer Information
Sequoia Capital Report
Goldman Sachs report

Links

Twitter: @kingnegritude
Linkedin: linkedin.com/in/ngozi-harrison
Website: kingnegritude.cargo.site