Abstract
We integrate foundational theories of meaning with a mathematical formalism of artificial general intelligence (AGI) to offer a comprehensive mechanistic explanation of meaning, communication, and symbol emergence. This synthesis holds significance for both AGI and broader debates concerning the nature of language, as it unifies pragmatics, logical truth conditional semantics, Peircean semiotics, and a computable model of enactive cognition, addressing phenomena that have traditionally evaded mechanistic explanation. By examining the conditions under which a machine can generate meaningful utterances or comprehend human meaning, we establish that the current generation of language models do not possess the same understanding of meaning as humans nor intend any meaning that we might attribute to their responses. To address this, we propose simulating human feelings and optimising models to construct weak representations. Our findings shed light on the relationship between meaning and intelligence, and how we can build machines that comprehend and intend meaning.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
We note that Grice later expanded upon the notion of m-intent [4, 5], and that there are other widely accepted descriptions of meaning (Russell, Frege, Searle, Davidson, Wittgenstein, Lewis, Kripke etc.), some of which we touch upon as part of our formalism. However, paper length limits what we discuss.
- 2.
The vocabulary \(\mathfrak {v}\) we single out represents the sensorimotor circuitry with which an organism enacts cognition - their brain, body, local environment and so forth.
- 3.
e.g. \(Z_s\) is the extension of s.
- 4.
Mind, body, local environment etc.
- 5.
The corresponding \(L_{\mathfrak {v}_\mathfrak {o}}\) is all sensorimotor activity in which the organism may engage.
- 6.
Note that this assumes qualia, preferences and so forth are part of physical reality, which means they are sets of declarative programs.
- 7.
A symbol system is every task to which one may generalise from one’s experiences. Only finitely many symbols may be entertained. In claiming our formalism pertains to meaning in natural language we are rejecting arguments, such as those of Block and Fodor [15], that a human can entertain an infinity of propositions (because time and memory are assumed to be finite, which is why \(\mathfrak {v}_\mathfrak {o}\) is finite).
- 8.
How an organism responds to a sign that means nothing is beyond this paper’s scope.
- 9.
Members of a species tend to have similar feelings, experiences and thus preferences.
- 10.
Albeit with some preference for simplicity imparted by regularisation.
References
Bennett, M.T.: Technical Appendices. Version 1.2.1 (2023). https://doi.org/10.5281/zenodo.7641742. github.com/ViscousLemming/Technical-Appendices
Grice, P.: Meaning. Philos. Rev. 66(3), 377 (1957)
Speaks, J.: Theories of meaning. In: The Stanford Encyclopedia of Philosophy. Spring 2021. Stanford University (2021)
Grice, P.: Utterers meaning and intention. Philos. Rev. 78(2), 147–177 (1969)
Grice, H.P.: Studies in the Way of Words. Harvard University Press, Cambridge (2007)
Bennett, M.T.: Symbol emergence and the solutions to any task. In: Goertzel, B., Iklé, M., Potapov, A. (eds.) AGI 2021. LNCS (LNAI), vol. 13154, pp. 30–40. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-93758-4_4
Bennett, M.T.: Compression, the fermi paradox and artificial super-intelligence. In: Goertzel, B., Iklé, M., Potapov, A. (eds.) AGI 2021. LNCS (LNAI), vol. 13154, pp. 41–44. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-93758-4_5
Bennett, M.T.: Computable artificial general intelligence (2022). https://arxiv.org/abs/2205.10513
Ward, D., Silverman, D., Villalobos, M.: Introduction: the varieties of enactivism. Topoi 36(3), 365–375 (2017). https://doi.org/10.1007/s11245-017-9484-6
Bennett, M.T., Maruyama, Y.: Philosophical specification of empathetic ethical artificial intelligence. IEEE Trans. Cogn. Dev. Syst. 14(2), 292–300 (2022)
Harnad, S.: The symbol grounding problem. Phys. D Nonlinear Phenom. 42(1), 335–346 (1990)
Howard, W.A.: The formulae-as-types notion of construction. Dedicated to H.B. Curry: Essays on Combinatory Logic, Lambda Calculus and Formalism. Ed. by J. Seldin and J. Hindley, pp. 479–490. Academic Press, Cambridge (1980)
Piccinini, G., Maley, C.: Computation in physical systems. In: The Stanford Encyclopedia of Philosophy. Summer. Stanford University (2021)
Atkin, A.: Peirces theory of signs. In: The Stanford Encyclopedia of Philosophy. Spring. Metaphysics Research Lab, Stanford University (2023)
Block, N., Fodor, J.: What psychological states are not. Philos. Rev. 81(2), 159–181 (1972)
Chalmers, D.: Facing up to the problem of consciousness. J. Conscious. Stud. 2(3), 200–219 (1995)
Boltuc, P.: The engineering thesis in machine consciousness. Techné Res. Philos. Technol. 16(2), 187–207 (2012)
Alexander, S.A.: The Archimedean trap: why traditional reinforcement learning will probably not yield AGI. J. Artif. Gen. Intell. 11(1), 70–85 (2020)
Gupta, A.: Definitions. In: The Stanford Encyclopedia of Philosophy. Winter 2021. Stanford University (2021)
Alexander, S.A., et al.: Extending environments to measure self-reflection in reinforcement learning. J. Artif. Gen. Intell. 13(1), 1–24 (2022)
Weizenbaum, J.: Computer Power and Human Reason: From Judgment to Calculation. W. H. Freeman & Co. (1976)
Floridi, L., Chiriatti, M.: GPT-3: its nature, scope, limits, and consequences. Minds Mach. 30, 681–694 (2020). https://doi.org/10.1007/s11023-020-09548-1
Acknowledgement
Appendices available on GitHub [1], supported by JST (JPMJMS2033).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Bennett, M.T. (2023). On the Computation of Meaning, Language Models and Incomprehensible Horrors. In: Hammer, P., Alirezaie, M., Strannegård, C. (eds) Artificial General Intelligence. AGI 2023. Lecture Notes in Computer Science(), vol 13921. Springer, Cham. https://doi.org/10.1007/978-3-031-33469-6_4
Download citation
DOI: https://doi.org/10.1007/978-3-031-33469-6_4
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-33468-9
Online ISBN: 978-3-031-33469-6
eBook Packages: Computer ScienceComputer Science (R0)