Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Cake that is Intelligence and Who Gets to Bake it: An AI Analogy and its Implications for Participation

Martin Mundt, Anaelia Ovalle, Felix Friedrich, A Pranav, Subarnaduti Paul, Manuel Brack, Kristian Kersting, William Agnew

TL;DR

These authors expand the AI cake analogy beyond a structural metaphor to the full AI life-cycle, linking data provenance, recipe design, training, evaluation, and distribution to social and ethical implications. They analyze core technical underpinnings—non-i.i.d. data, homogenized model architectures, costly continual learning, and proxy-based evaluation—that shape real-world outcomes. The paper offers cross-disciplinary, actionable recommendations at each lifecycle stage to promote participation, transparency, and sustainable AI design, while warning against ethics dumping and technosolutionism. Overall, the work provides a framework to align technical development with social accountability and invites broader collaboration among researchers, practitioners, and policymakers.

Abstract

In a widely popular analogy by Turing Award Laureate Yann LeCun, machine intelligence has been compared to cake - where unsupervised learning forms the base, supervised learning adds the icing, and reinforcement learning is the cherry on top. We expand this 'cake that is intelligence' analogy from a simple structural metaphor to the full life-cycle of AI systems, extending it to sourcing of ingredients (data), conception of recipes (instructions), the baking process (training), and the tasting and selling of the cake (evaluation and distribution). Leveraging our re-conceptualization, we describe each step's entailed social ramifications and how they are bounded by statistical assumptions within machine learning. Whereas these technical foundations and social impacts are deeply intertwined, they are often studied in isolation, creating barriers that restrict meaningful participation. Our re-conceptualization paves the way to bridge this gap by mapping where technical foundations interact with social outcomes, highlighting opportunities for cross-disciplinary dialogue. Finally, we conclude with actionable recommendations at each stage of the metaphorical AI cake's life-cycle, empowering prospective AI practitioners, users, and researchers, with increased awareness and ability to engage in broader AI discourse.

The Cake that is Intelligence and Who Gets to Bake it: An AI Analogy and its Implications for Participation

TL;DR

These authors expand the AI cake analogy beyond a structural metaphor to the full AI life-cycle, linking data provenance, recipe design, training, evaluation, and distribution to social and ethical implications. They analyze core technical underpinnings—non-i.i.d. data, homogenized model architectures, costly continual learning, and proxy-based evaluation—that shape real-world outcomes. The paper offers cross-disciplinary, actionable recommendations at each lifecycle stage to promote participation, transparency, and sustainable AI design, while warning against ethics dumping and technosolutionism. Overall, the work provides a framework to align technical development with social accountability and invites broader collaboration among researchers, practitioners, and policymakers.

Abstract

In a widely popular analogy by Turing Award Laureate Yann LeCun, machine intelligence has been compared to cake - where unsupervised learning forms the base, supervised learning adds the icing, and reinforcement learning is the cherry on top. We expand this 'cake that is intelligence' analogy from a simple structural metaphor to the full life-cycle of AI systems, extending it to sourcing of ingredients (data), conception of recipes (instructions), the baking process (training), and the tasting and selling of the cake (evaluation and distribution). Leveraging our re-conceptualization, we describe each step's entailed social ramifications and how they are bounded by statistical assumptions within machine learning. Whereas these technical foundations and social impacts are deeply intertwined, they are often studied in isolation, creating barriers that restrict meaningful participation. Our re-conceptualization paves the way to bridge this gap by mapping where technical foundations interact with social outcomes, highlighting opportunities for cross-disciplinary dialogue. Finally, we conclude with actionable recommendations at each stage of the metaphorical AI cake's life-cycle, empowering prospective AI practitioners, users, and researchers, with increased awareness and ability to engage in broader AI discourse.

Paper Structure

This paper contains 15 sections, 7 equations, 2 figures.

Table of Contents

  1. Introduction
  2. AI cake: an accurate analogy
  3. Ingredients and their origin
  4. The "best" cake recipes
  5. Cake foundation and the baking process
  6. What makes for a tasty cake?
  7. "If they do not have bread, let them eat cake!" --- Sharing and (over-)selling
  8. Technological underpinnings: Why it is difficult to change the AI cake
  9. "Ingredients": the challenge of data "i.i.d.-ness"
  10. "Recipes": Homogenization of instructions
  11. "Baking process": the interference dilemma
  12. "Taste": limits of averages and aggregates
  13. "(Over-)Selling": the abundant surrogate impasse
  14. Limitations and Disclaimer
  15. Conclusion

Figures (2)

  • Figure 1: An illustration of the AI--cake analogy. (Left) Traditionally, the cake was used to provide a structural metaphor for machine intelligence, relating unsupervised (bulk), supervised (icing), and reinforcement learning (cherry) paradigms. (Right) Our re-conceptualized analogy extends the original metaphor by drawing parallels to the way AI's ingredients are sourced, recipes crafted, and ultimately how the metaphorical cake is baked, tasted and sold.
  • Figure 2: An overview of section three's considerations with respect to the AI-cake's technical foundation and our future recommendations for each process stage from ingredients and recipes, to baking, tasting and selling.