Real Safety AI Foundation / Research

Machine Cognition, Consciousness, and Moral Status

The Communities Are Not Edge Cases: A Standing-Based Definition of Artificial General Intelligence

Publisher: Real Safety AI Foundation

Working paper. Not peer reviewed.

Written
May 2026
Version
v10
Pages
22
Download PDFPlain language slides (PDF)Export BibTeXExport RIS

Abstract

The contemporary conversation about Artificial General Intelligence operates without a working definition. Major laboratories use the term as a fundraising milestone, a regulatory framing device, and a release marker for whatever capability the laboratory has most recently shipped. The resulting ambiguity is functionally a control mechanism: a definition that stays vague cannot be regulated, audited, or held to a public standard. This paper proposes a standing-based, two-axis definition of AGI that extends Chollet’s (2019) skill-acquisition-efficiency framework into the domains of harm recognition and ethical calibration. The definition treats AGI as the union of human capability, the ability to do what any human can do at expert level when operating in a domain, distinct from superintelligence, which exceeds human capability across domains. Because that union includes the domains whose experts study culture, disability, race, and bias, recognizing harm to a community is not an add-on to intelligence but one of the domains the definition already covers. AGI is recognized at the threshold where a system can be applied to any domain without the human first having to teach it that domain, evaluated against substrates that carry community-specific harm standards and ethical frameworks, with recognition authority allocated to the affected standing vantage points rather than to the developer. The Harm Blindness Framework, a 312-case corpus of historical and corporate harms across cultures and centuries, is offered as an existence proof that such a harm substrate is buildable and operable, not as a complete one. The ethics substrate is the next required build. The paper grounds the two-axis definition in the civil rights and disability rights tradition, the universal design methodology developed by Mace and colleagues at North Carolina State University, and the philosophy of language tradition’s distinction between communication and understanding. The communities the AGI conversation treats as edge cases are placed at the center of the evaluation, because the center of the evaluation is wherever the recognition operation is hardest. The definition forces public, auditable substrates; forces engagement with civil rights and disability rights doctrine; and forces a structural choice for design that the disability community formalized four decades ago: design for everyone or design for no one.

Keywords

  • AGI
  • standing-based definition
  • substrate-based testing
  • Harm Blindness Framework
  • universal design
  • ethical pluralism
  • philosophy of language
  • communication
  • disability studies
  • disability rights
  • civil rights
  • disparate impact

Plain language slides

First slide of the plain language summary of The Communities Are Not Edge Cases: A Standing-Based Definition of Artificial General IntelligenceOpen the 23-slide summary (PDF)

Suggested citation

Gilly, Travis. "The Communities Are Not Edge Cases: A Standing-Based Definition of Artificial General Intelligence." Real Safety AI Foundation Working Paper, May 2026. https://realsafetyai.org/research/communities-are-not-edge-cases/

References (42)

  1. Appiah, K. A. (2006). Cosmopolitanism: Ethics in a World of Strangers. W. W. Norton.
  2. Bai, Y., Kadavath, S., Kundu, S., et al. (2022). Constitutional AI: Harmlessness from AI feedback. arXiv preprint arXiv:2212.08073. https://arxiv.org/abs/2212.08073
  3. Barocas, S., & Selbst, A. D. (2016). Big Data’s disparate impact. California Law Review, 104(3), 671–732. https://doi.org/10.15779/Z38BG31
  4. Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic parrots: Can language models be too big? Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’21), 610–623. https://doi.org/10.1145/3442188.3445922
  5. Benjamin, R. (2019). Race After Technology: Abolitionist Tools for the New Jim Code. Polity Press.
  6. Bostrom, N. (2014). Superintelligence: Paths, Dangers, Strategies. Oxford University Press.
  7. Buolamwini, J., & Gebru, T. (2018). Gender Shades: Intersectional accuracy disparities in commercial gender classification. Proceedings of the 1st Conference on Fairness, Accountability and Transparency, in Proceedings of Machine Learning Research, 81, 77–91.
  8. Charlton, J. I. (1998). Nothing About Us Without Us: Disability Oppression and Empowerment. University of California Press.
  9. Chen, E. K., Belkin, M., Bergen, L., & Danks, D. (2026). Does AI already have human-level intelligence? The evidence is clear [Comment]. Nature. https://doi.org/10.1038/d41586-026-00285-6
  10. Chollet, F. (2019). On the measure of intelligence. arXiv preprint arXiv:1911.01547. https://arxiv.org/abs/1911.01547
  11. Connell, B. R., Jones, M., Mace, R., Mueller, J., Mullick, A., Ostroff, E., Sanford, J., Steinfeld, E., Story, M., & Vanderheiden, G. (1997). The principles of universal design (Version 2.0). North Carolina State University, The Center for Universal Design.
  12. Coulthard, G. S. (2014). Red Skin, White Masks: Rejecting the Colonial Politics of Recognition. University of Minnesota Press.
  13. Eubanks, V. (2018). Automating Inequality: How High-Tech Tools Profile, Police, and Punish the Poor. St. Martin’s Press. ISBN 978-1-250-07431-7.
  14. Gumperz, J. J. (1982). Discourse Strategies. Cambridge University Press.
  15. Hymes, D. (1972). Models of the interaction of language and social life. In J. J. Gumperz & D. Hymes (Eds.), Directions in Sociolinguistics: The Ethnography of Communication (pp. 35–71). Holt, Rinehart and Winston.
  16. Mace, R. L. (1985, November). Universal design: Barrier free environments for everyone. Designers West, 33(1), 147–152.
  17. MacIntyre, A. (1981). After Virtue: A Study in Moral Theory. University of Notre Dame Press.
  18. Macklin, R. (1999). Against Relativism: Cultural Diversity and the Search for Ethical Universals in Medicine. Oxford University Press.
  19. Mingus, M. (2017, April 12). Access intimacy, interdependence and disability justice. Leaving Evidence [blog]. https://leavingevidence.wordpress.com/2017/04/12/access-intimacy-interdependence-and-disability-justice/
  20. Noble, S. U. (2018). Algorithms of Oppression: How Search Engines Reinforce Racism. NYU Press.
  21. Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019). Dissecting racial bias in an algorithm used to manage the health of populations. Science, 366(6464), 447–453. https://doi.org/10.1126/science.aax2342
  22. O’Neil, C. (2016). Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown.
  23. Pickering, M. J., & Garrod, S. (2004). Toward a mechanistic psychology of dialogue. Behavioral and Brain Sciences, 27(2), 169–190. https://doi.org/10.1017/S0140525X04000056
  24. Rothstein, R. (2017). The Color of Law: A Forgotten History of How Our Government Segregated America. Liveright Publishing. ISBN 978-1-63149-285-3.
  25. Sen, A. (2009). The Idea of Justice. Belknap Press of Harvard University Press.
  26. Strathern, M. (1997). “Improving ratings”: Audit in the British university system. European Review, 5(3), 305–321. https://doi.org/10.1017/S1062798700002660
  27. Suchman, L. A. (1987). Plans and Situated Actions: The Problem of Human-Machine Communication. Cambridge University Press.
  28. Tuck, E., & Yang, K. W. (2012). Decolonization is not a metaphor. Decolonization: Indigeneity, Education & Society, 1(1), 1–40.
  29. Walzer, M. (1983). Spheres of Justice: A Defense of Pluralism and Equality. Basic Books.
  30. Whyte, K. P. (2018). Settler colonialism, ecology, and environmental injustice. Environment and Society, 9(1), 125–144. https://doi.org/10.3167/ares.2018.090109
  31. Wong, D. B. (2006). Natural Moralities: A Defense of Pluralistic Relativism. Oxford University Press.
  32. Yampolskiy, R. V. (2021). On the differences between human and machine intelligence. CEUR Workshop Proceedings, Vol. 2916. https://arxiv.org/abs/2007.07710

Legal authorities and primary sources

  1. Garland v. Cargill, 602 U.S. 406 (2024).
  2. Griggs v. Duke Power Co., 401 U.S. 424 (1971).
  3. Jacobellis v. Ohio, 378 U.S. 184 (1964).
  4. Americans with Disabilities Act of 1990, 42 U.S.C. §12101 et seq.
  5. Rehabilitation Act of 1973, Section 504, 29 U.S.C. §794.
  6. Civil Rights Act of 1964, Title VI & Title VII.
  7. Federal Rules of Civil Procedure, Rule 23(b)(2).
  8. 34 C.F.R. Part 104 [Section 504 implementing regulations].
  9. United Nations Convention on the Rights of Persons with Disabilities, opened for signature 30 March 2007, 2515 U.N.T.S. 3, Article 4(3).
  10. OpenAI, “Our Charter,” https://openai.com/charter/ (last visited May 2026).

All research