A predictive human model of language challenges traditional views in linguistics and pretrained transformer research DOI Creative Commons
Sergio Torres–Martínez

Language and Semiotic Studies, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 21, 2024

Abstract This paper introduces a theory of mind that positions language as cognitive tool in its own right for the optimization biological fitness. I argue human reconstruction reality results from memory and adaptation to uncertain environmental conditions reaffirmation Self-as-symbol. demonstrate pretrained models, such ChatGPT, lack embodied grounding, which compromises their ability adequately model world through due absence subjecthood conscious states event recognition partition. At deep level, challenge notion constitution semiotic Self relies on computational reflection , arguing against reducing representation data structures emphasizing importance positing accurate models language. underscores distinction between transformers posthuman agents humans purposeful agents, emphasizes capacity adjustment optimization. One main conclusions this is integrate information does not amount phenomenal consciousness argued by Information Integration Theory . Moreover, while exhibit superior capacity, they real providing them with multiscalar experience anchored physical world, characteristic cognition. However, anticipates emergence new silico conceptualizers capable defining themselves symbolic contours specific goals.

Язык: Английский

Semiosic Translation: a Bayesian-heuristic theory of translation and translating DOI Creative Commons
Sergio Torres–Martínez

Language and Semiotic Studies, Год журнала: 2024, Номер 10(2), С. 167 - 202

Опубликована: Фев. 29, 2024

Abstract In the context of Semiosic Translation , two elements are essential for a translation to emerge: body–brain–context interface (extended mind) and sign systems making up output. this paper, I explain how renewed view body as Bayesian-heuristic Semiotic Prior helps understand in more holistic manner motivations agentive character translation, defined herein phenomenological grasp world. Central present proposal is idea that bodily self-stabilization (homeostasis) brain-driven correction (allostasis) provide translator-agents with maps action upon world semiotic nature. All occurs thanks information weighing (Bayesian) cue-driven (heuristic) types inference whereby exteroceptive (exogenous) interoceptive (inner-body) signals converge create sense awareness responsible construction symbolic persona (the translator-agent).

Язык: Английский

Процитировано

4
Dehumanizing the human, humanizing the machine: organic consciousness as a hallmark of the persistence of the human against the backdrop of artificial intelligence DOI
Sergio Torres–Martínez

AI & Society, Год журнала: 2025, Номер unknown

Опубликована: Янв. 28, 2025

Язык: Английский

Процитировано

0
Embodied essentialism in the reconstruction of the animal sign in robot animal design DOI
Sergio Torres–Martínez

Biosystems, Год журнала: 2024, Номер 238, С. 105178 - 105178

Опубликована: Март 13, 2024

Язык: Английский

Процитировано

2
A predictive human model of language challenges traditional views in linguistics and pretrained transformer research DOI Creative Commons
Sergio Torres–Martínez

Language and Semiotic Studies, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 21, 2024

Abstract This paper introduces a theory of mind that positions language as cognitive tool in its own right for the optimization biological fitness. I argue human reconstruction reality results from memory and adaptation to uncertain environmental conditions reaffirmation Self-as-symbol. demonstrate pretrained models, such ChatGPT, lack embodied grounding, which compromises their ability adequately model world through due absence subjecthood conscious states event recognition partition. At deep level, challenge notion constitution semiotic Self relies on computational reflection , arguing against reducing representation data structures emphasizing importance positing accurate models language. underscores distinction between transformers posthuman agents humans purposeful agents, emphasizes capacity adjustment optimization. One main conclusions this is integrate information does not amount phenomenal consciousness argued by Information Integration Theory . Moreover, while exhibit superior capacity, they real providing them with multiscalar experience anchored physical world, characteristic cognition. However, anticipates emergence new silico conceptualizers capable defining themselves symbolic contours specific goals.

Язык: Английский

Процитировано

1