Between world models and model worlds: on generality, agency, and worlding in machine learning

Research output: Journal contributionsJournal articlesResearchpeer-review

Authors

The article offers a discursive account of what generality in machine learning research means and how it is constructed in the development of general artificial intelligence from the perspectives of cultural and media studies. I discuss several technical papers that outline novel architectures in machine learning and how they conceive of the “world”. The agency to learn and the learning curriculum are modulated through worlding (in the sense of setting up and unfolding of the world for artificial agents) in machine learning engineering. In recent computer science articles, large models trained on Internet-scale datasets are framed as general world simulators—despite their partiality, historicity, finite nature, and cultural specificity. I introduce the notion of “model worlds” to refer to composable interactive environments designed for the purpose of machine learning that partake in legitimising that claim. I discuss how large models are grounded through interaction in model worlds, arguing that model worlds mediate between the sheer scale of language models and their hypothetical capacity to generalise to new tasks and domains, rehashing the empiricist logic of “big data”. Further, I show that the emerging capacity of artificial agents to generalise redraws the epistemic boundary between artificial agents and their learning environments. Consequently, superficial statistics of language models and abstract action are made meaningful in distilled model worlds, giving rise to synthetic agency.

Original languageEnglish
JournalAI and Society
Number of pages13
ISSN0951-5666
DOIs
Publication statusE-pub ahead of print - 07.10.2024

Bibliographical note

Publisher Copyright:
© The Author(s) 2024.

Recently viewed

Publications

  1. The effectiveness of nudging
  2. Vorwort
  3. The impact of emotions, moods, and other affect-related variables on creativity, innovation and initiative
  4. Characterization of the Basic Types of Lunar Highland Breccias by Quantitative Textural Analysis
  5. Dis/Ability and Digital Cultures. A Media-Archaeological Perspective on Inclusion as a Cipher
  6. The Porter Hypothesis Revisited
  7. Counts of all walks as atomic and molecular descriptors.
  8. Classification of playing position in elite junior Australian football using technical skill indicators
  9. Challenges and Choices in Conducting Qualitative Meta-studies
  10. Transformations of pesticides in the atmosphere
  11. Planning for Sea Spaces II
  12. Numerical simulation of friction extrusion
  13. Deciding whether to work after retirement
  14. Energy transitions and national development indicators
  15. Organizational error management culture and its impact on performance: a two-study replication
  16. Irish English and Variational Pragmatics
  17. A Reference Model for Data-driven Business Model Innovation Initiatives in Incumbent Firms
  18. It Is Belief in Dualism, and Not Free Will, That Best Predicts Helping: A Conceptual Replication and Extension of Baumeister et al. (2009)
  19. Experimental and numerical study of laser shock peening process of AA2198-T3 and AA2198-T8
  20. Pricing effects when competitors arrive
  21. Adapting Growth Models for Digital Startups
  22. What makes online professional development effective?
  23. Mindfulness at work
  24. How self-regulation helps to master negotiation challenges
  25. Model predictive control of transistor pulse converter for feeding electromagnetic valve actuator with energy storage
  26. Liveness Formats
  27. Digital Games and Fan-Discourse
  28. Towards a critical understanding of work in ecological economics
  29. Frühe Medienbildung
  30. Designing instructional technology from an emotional perspective
  31. Review