L’ITALIANO DELLE INTELLIGENZE ARTIFICIALI GENERATIVE
Open accessThe DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index.
Abstract
I chatbot basati su modelli linguistici di grandi dimensioni sono un tipo di intelligenza artificiale in grado di interagire con gli utenti rispondendo con un testo coerente e contestualmente rilevante a un ampio numero di richieste. Si tratta di una tecnologia in rapido sviluppo, che promette di cambiare il modo in cui alcuni tipi di informazioni sono confezionati e diffusi: sempre più spesso ci si troverà a leggere un testo composto da una macchina. Questo articolo evidenzia alcuni aspetti essenziali dell’italiano delle intelligenze artificiali generative, attraverso l’analisi dei testi prodotti dai principali chatbot disponibili nella nostra lingua: ChatGPT, Bard e Bing Chat. In particolare, dopo aver sondato il ruolo dell’italiano nei database che permettono il loro funzionamento, l’attenzione è stata rivolta alle strategie comunicative destinate a simulare una personalità umana e alle caratteristiche linguistiche e stilistiche di alcuni testi di argomento scientifico composti dai chatbot. I risultati mostrano la capacità dei modelli linguistici di riprodurre i tratti più tipici dei testi che sono chiamati a emulare; tuttavia, sono emersi anche importanti limiti dal punto di vista informativo e comunicativo. Il confronto con le risposte composte in lingua inglese ha dimostrato come alcuni di questi difetti siano correlati alla qualità dei corpora di addestramento. Altri, però, sembrano effetti indesiderati di quella stessa struttura statistica che rende il funzionamento delle intelligenze artificiali possibile. The Italian of Generative Artificial Intelligences Chatbots based on large language models are a type of artificial intelligence capable of interacting with users by providing coherent and contextually relevant text responses to a wide range of requests. This is a rapidly advancing technology which promises to change the way certain types of information are generated and transmitted: it will soon become increasingly common to encounter machine-generated text. This article highlights the essential aspects of the Italian used by generative artificial intelligences, through the analysis of the texts produced by leading chatbots available in the language: ChatGPT, Bard and Bing Chat. Specifically, after examining the role of Italian in the databases enabling their operation, attention is directed toward communicative strategies employed to simulate a human personality and on the linguistic and stylistic characteristics of some scientific texts composed by chatbots. The results demonstrate the models' abilities to replicate the most typical features of the texts they are tasked to emulate. Nonetheless, significant limitations have also emerged from an informational and communicative perspective. A comparison with responses generated in English has revealed that some of these shortcomings are linked to the quality of the training corpora. Others, however, appear to be unintended consequences of the same statistical structure that makes the functioning of artificial intelligences possible.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
Chain-Of-Thought Prompting Elicits Reasoning in Large Language Models
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q. V., & Zhou, D. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1800 · Cited by 4,505
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q., & Zhou, D. (2022) · arXiv · doi:10.48550/arxiv.2201.11903 · Cited by 4,263
Assessing Acceptance of Assistive Social Agent Technology by Older Adults: the Almere Model
Heerink, M., Kröse, B., Evers, V., & Wielinga, B. (2010) · International Journal of Social Robotics · doi:10.1007/s12369-010-0068-5 · Cited by 1,229
Pelau, C., Dabija, D.-C., & Ene, I. (2021) · Computers in Human Behavior · doi:10.1016/j.chb.2021.106855 · Cited by 889
A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT
White, J., Fu, Q., Hays, S., Sandborn, M., Olea, C., Gilbert, H., Elnashar, A., Spencer-Smith, J., & Schmidt, D. C. (2023) · arXiv · doi:10.48550/arxiv.2302.11382 · Cited by 838
In the shades of the uncanny valley: An experimental study of human–chatbot interaction
Ciechanowski, L., Przegalinska, A., Magnuski, M., & Gloor, P. (2019) · Future Generation Computer Systems · doi:10.1016/j.future.2018.01.055 · Cited by 651
Can computer personalities be human personalities?
Nass, C., Moon, Y., Fogg, B. J., Reeves, B., & Dryer, D. C. (1995) · International Journal of Human-Computer Studies · doi:10.1006/ijhc.1995.1042 · Cited by 615
In AI We Trust: Ethics, Artificial Intelligence, and Reliability
Ryan, M. (2020) · Science and Engineering Ethics · doi:10.1007/s11948-020-00228-y · Cited by 600
Anthropomorphism: Opportunities and Challenges in Human–Robot Interaction
Złotowski, J., Proudfoot, D., Yogeeswaran, K., & Bartneck, C. (2015) · International Journal of Social Robotics · doi:10.1007/s12369-014-0267-6 · Cited by 598
AI as Agency Without Intelligence: on ChatGPT, Large Language Models, and Other Generative Models
Floridi, L. (2023) · Philosophy & Technology · doi:10.1007/s13347-023-00621-y · Cited by 404
Gender and Representation Bias in GPT-3 Generated Stories
Lucy, L., & Bamman, D. (2021) · Proceedings of the Third Workshop on Narrative Understanding · doi:10.18653/v1/2021.nuse-1.5 · Cited by 287
Quantifying Memorization Across Neural Language Models
Carlini, N., Ippolito, D., Jagielski, M., Lee, K., Tramer, F., & Zhang, C. (2022) · arXiv · doi:10.48550/arxiv.2202.07646 · Cited by 204
Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4Preprint
Chang, K. K., Cramer, M., Soni, S., et al. (2023) · Conference on Empirical Methods in Natural Language Processing · doi:10.48550/arxiv.2305.00118 · Cited by 204
How does anthropomorphism improve human-AI interaction satisfaction: a dual-path model
Xie, Y., Zhu, K., Zhou, P., & Liang, C. (2023) · Computers in Human Behavior · doi:10.1016/j.chb.2023.107878 · Cited by 150
Training Data Extraction From Pre-trained Language Models: A Survey
Ishihara, S. (2023) · Workshop on Trustworthy Natural Language Processing (TrustNLP) · doi:10.18653/v1/2023.trustnlp-1.23 · Cited by 84
Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4
Chang, K., Cramer, M., Soni, S., & Bamman, D. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.453 · Cited by 83
Training Data Extraction From Pre-trained Language Models: A SurveyPreprint
Ishihara, S. (2023) · TRUSTNLP · doi:10.48550/arxiv.2305.16157 · Cited by 66
Caffi, C. (2006) · Elsevier eBooks · doi:10.1016/b0-08-044854-2/00392-8 · Cited by 61
Making and Using AI in the Library: Creating a BERT Model at the National Library of Sweden
Haffenden, C., Fano, E., Malmsten, M., & Börjeson, L. (2023) · College & Research Libraries · doi:10.5860/crl.84.1.30 · Cited by 59
A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT
(2024) · Proceedings of the 30th Conference on Pattern Languages of Programs · doi:10.64346/plop2023p05 · Cited by 32
Scaglia, C. (2003) · Aisberg (University of Bergamo) · doi:10.6092/lef_16_p109 · Cited by 10
Capire la lingua della scienza
Dardano, M. (2008) · Iris (Roma Tre University) · Cited by 1
Extracting Training Data from Diffusion Models
Carlini, N., Hayes, J., Nasr, M., Jagielski, M., Sehwag, V., Tramèr, F., Balle, B., Ippolito, D., & Wallace, E. (2023) · arXiv · doi:10.48550/arxiv.2301.13188
Research works printed but not linked: 6
- Linguaggi scientifici e intersezioni tra letteratura e scienza. Atti della giornata di studi tenuta il 25 marzo 1998 presso l’IPSIA “Europa” di Roma, Centro studi “Giuseppe Gioacchino Belli”, Comune di Roma, VI circoscrizione, Ufficio Cultura, Roma, pp. 31-44.
- Dardano M. (1994), “I linguaggi scientifici”, in Serianni L., Trifone P. (a cura di), Storia della lingua parlata, vol. II, Einaudi, Torino, pp. 497-551.
- Fedriani C. (2020), “La mitigazione degli atti richiestivi: variazione e mutamento nella storia dell’italiano”, in Alfieri G., Alfonzetti G., Motta D., Sardo R. (2020), pp. 565-573.
- Mortara Garavelli B. (1993), “Strutture testuali e retoriche”, in Sobrero A.A. (a cura di), Introduzione all’italiano contemporaneo. Le strutture, Laterza, Roma-Bari, pp. 371-404.
- Reeves B., Nass C. (1996), The media equation: How people treat computers, television, and new media like real people and places, Cambridge University Press, Cambridge.
- Sobrero A.A. (1993), “Lingue speciali”, in Sobrero A.A. (a cura di), Introduzione all’italiano contemporaneo. Le strutture, Laterza, Roma-Bari, pp. 237-277.
Cited by
Valutare la qualità dei testi generati in lingua italiana
Tavosanis, M. L. A. (2024) · AI-Linguistica Linguistic Studies on AI-Generated Texts and Discourses · doi:10.62408/ai-ling.v1i1.14 · Cited by 14
Cesare, A.-M. D. (2024) · Lingue e culture dei media · doi:10.54103/2532-1803/22388 · Cited by 10
Esercizi di stile. La narrativa per l’infanzia delle intelligenze artificiali
Cicero, F. (2025) · AI-Linguistica Linguistic Studies on AI-Generated Texts and Discourses · doi:10.62408/ai-ling.v2i2.19 · Cited by 6
“L’impronta dell’inglese e il ridotto ventaglio espressivo sono una cifra stilistica anche dei testi di divulgazione scientifica dell’IA (cfr. Cicero 2023 e 2024).”
Self-citationChatGPT for President! Presupposed content in politicians versus GPT-generated texts
Garassino, D., Brocca, N., & Masia, V. (2025) · Applied Corpus Linguistics · doi:10.1016/j.acorp.2025.100156 · Cited by 2
“In this study, we present the results of a comparative analysis be-tween political speeches produced by politicians and by GPT-4 (OpenAI, 2024).”
InfluentialTavosanis, M. L. A. (2025) · AI-Linguistica Linguistic Studies on AI-Generated Texts and Discourses · doi:10.62408/ai-ling.v2i2.23 · Cited by 2
“…intelligenze artificiali generative mostrano una capacità notevole di produrre testi rispettando le regole grammaticali di una lingua, soprattutto nelle varietà più formali (per la situazione generale riguardo l’italiano : Cicero 2023; Tavosanis 2024; Antonelli 2025 per una prospettiva diacronica).”
Metadata
- Authors
- Cicero, F.
- Affiliations
- University of Naples - L'Orientale (IT)
- Corresponding author
- Cicero, Francesco
- ORCID
- Cicero, Francesco
- Venue
- Italiano LinguaDue
- Publisher
- Milano University Press
- Published
- 2023-12-15
- Volume
- 15
- Issue
- 2
- Pages
- 733-761
- ISSN
- 2037-3597 (ISSN-L)
- OpenAlex
- W4389794060
- Cited by
- 5
- References
- 22
- Citations by year
- 2026: 1 · 2025: 2 · 2024: 2
- FWCI
- 0.44
- Type
- journal article
- Language
- it
- Open access
- diamond
- doi.org
- Licence
- Locations
- Italiano LinguaDue (publishedVersion, cc-by-sa)
- Archivio Istituzionale della Ricerca (Universita Degli Studi Di Milano) (submittedVersion, other-oa)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Riviste UNIMI (Università degli studi di Milano) (submittedVersion, cc-by-nc-nd)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Social Sciences › Arts and Humanities › Language and Linguistics › Linguistic research and analysis
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- generative AI, chatbots, large language models, Italian language, ChatGPT, Google Bard, Bing Chat, AI-generated text, scientific texts, communication strategies, training corpus, stylistic features, Italian as a second language, Italian as a foreign language
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.