Page 43 - Tvorjenje pomena in večrazsežna – večkodna pismenost

Ernest Ženko Generative Artificial Intelligence and the Question
University of Primorska, Slovenia of Meaning: A Semiological Perspective from Saussure
ernest.zenko@upr.si
© 2026 Ernest Ženko
Meaning-Making, Multiliteracies
and Multimodality
Abstracts of the International
Symposium
Koper, 19–20 March 2026
to Lacan
This presentation approaches generative artificial intelligence (AI) as a boundary
case of a semiological perspective that radicalizes structural theories of the sign
while simultaneously opening up fundamental questions concerning the subject
of meaning (Mitchell, 2019). The point of departure is Saussure’s (1997) concep-
tion of the sign as a differential relation between signifier and signified, in which
meaning is not anchored in reference but grounded in a system of differences.
Large language models and multimodal models that underpin generative AI not
only confirm this understanding but push it to an extreme, insofar as signs in this
context emerge exclusively within a formal system of relations.
In a second step, the analysis draws on Barthes’s (2015, 2021) theory of polysemy,
myth, and the ‘death of the author.’ Within this framework, generative AI can be
understood as a technical realisation of Barthes’s thesis: the author withdraws as
the source of meaning, and their place is taken by an algorithmic process oper-
ating at the level of signifiers within a discursively heterogeneous archive. The
multimodality of generated signs does not arise from deliberate semantic inten-
tion, but from the interweaving of different semiotic codes already present in the
corpus, which converge in each output as a crossroads of linguistic, visual, cul-
tural, and ideological registers.
Finally, the analysis turns to Lacanian (2010) psychoanalysis, in particular to the
notion of the symbolic order and the primacy of the signifier. In this sense, gen-
erative AI functions as a machine that operates exclusively within the symbolic
register: it produces chains of signifiers without access to the imaginary body
and without a relation to the real. AI does not »understand« meaning; rather, it
produces meaning as an effect of the sliding of signifiers.
This perspective allows generative AI to be understood as a system without an
unconscious – or, more precisely, as a system that reproduces the formal charac-
teristics of the symbolic order without assuming the position of the subject. Gen-
erative AI does not replace human meaning-making; instead, it places it in a new
relation, revealing the extent to which meaning is always already structured by
codes,discourses,andsymbolicrelations.Multimodalitythusemergesasafunda-
mental feature of contemporary symbolic production, which becomes actualized
only in relation to human interpretation, responsibility, and critical reading.
Barthes, R. (2015). Mitologije. (V. Troha, Trans.). Krtina.
Barthes, R. (2021). Šelestenje jezika. (K. Rotar, Trans.). Krtina.
Lacan, J. (2010). Štirje temeljni koncepti psihoanalize. (R. Močnik, Z. Skušek, S. Žižek,
Trans.). Društvo za teoretsko psihoanalizo.
Mitchell, M. (2019). Artificial intelligence: A guide for thinking humans. Farrar, Straus
and Giroux.
Saussure, F. (1997). Predavanja iz splošnega jezikoslovja. (B.M.Turk, Trans.).ISH
Fakulteta za podiplomski humanistični študij.
https://doi.org/10.26493/978-961-293-565-8.40 43