Does ChatGPT-4 dream of offensive tweets? An approximation to the potential contributions of generative models in detecting illicit speeches

Article Sidebar

PDF (Castellano) ePUB (Castellano)

Main Article Content

Mario Santisteban Galarza
University of Basque Country / Euskal Herriko Unibertsitatea
https://orcid.org/0000-0002-6191-3044
Jesús C. Aguerri
Miguel Hernández University of Elche
https://orcid.org/0000-0002-7730-8527

The latest advances in generative artificial intelligence appear to allow artificial intelligence models to be equipped with capabilities as relevant to the legal field as to arguing their own decisions. This study approximates the capabilities of the ChatGPT-4 model in the context of the detection of hate speeches according to their conceptualization in article 510.1. a) of the Spanish Criminal Code. For this purpose, ChatGPT-4’s reasoning and decisions will be compared from proven factual accounts with the decisions of the bodies that judged the respective cases, studying the limits and potential of these systems in the legal field.

Keywords
artificial intelligence, hate speech, ChatGPT, generative artificial intelligence, fundamental rights

Article Details

How to Cite
Santisteban Galarza, Mario; and C. Aguerri, Jesús. “Does ChatGPT-4 dream of offensive tweets? An approximation to the potential contributions of generative models in detecting illicit speeches”. IDP. Internet, Law and Politics E-Journal, no. 39, pp. 1-12, doi:10.7238/idp.v0i39.416638.
Rights
Creative Commons License

This work is licensed under a Creative Commons Attribution-NoDerivatives 3.0 Unported License.

(c) Mario Santisteban Galarza, Jesús C. Aguerri, 2023


Copyright

Contents published in IDP are subject to a Creative Commons Attribution-No Derivative Works 3.0 Spain licence, the full text of which can be consulted on http://creativecommons.org/licenses/by-nd/3.0/es/deed.en.

Thus, they may be copied, distributed and broadcast provided that the author and IDP are cited, as shown in the recommended citation that accompanies each article. Derivative works are not permitted.

Authors are responsible for obtaining the necessary permission to use copyrighted images.

 

Assignment of intellectual property rights

The author non exclusively transfers the rights to use (reproduce, distribute, publicly broadcast or transform) and market the work, in full or part, to the journal’s editors in all present and future formats and modalities, in all languages, for the lifetime of the work and worldwide.

The author must declare that he is the original author of the work. The editors shall thus not be held responsible for any obligation or legal action that may derive from the work submitted in terms of violation of third parties’ rights, whether intellectual property, trade secret or any other right.

Author Biographies

Mario Santisteban Galarza, University of Basque Country / Euskal Herriko Unibertsitatea

Jurist, predoctoral investigator of the University of the Basque Country (UPV/EHU) in the Department of Law of the Company and Civil Law. His work focuses on the relationships between law and technology, particularly freedom of expression on the internet and the powers of digital platforms to control speech. 

Jesús C. Aguerri, Miguel Hernández University of Elche

Postdoctoral researcher (Juan de la Cierva-Training at the Crímina Center for the Study and Prevention of Crime). He has a PhD in Sociology from the University of Zaragoza and has developed his work around the sociology of cyberspace and freedom of expression management.

References

ALASTUEY DOBÓN, C. (2016). «Discurso del odio y negacionismo en la reforma del Código Penal de 2015». Revista Electrónica de Ciencia Penal y Criminología, vol. 18-14, págs.1-38 [en línea]. Disponible: http://criminet.ugr.es/recpc/18/recpc18-14.pdf

ANDRÉS PUEYO, A.; ARBACH-LUCIONI, K.; REDONDO, S. (2017). «The RisCanvi: A New Tool for Assessing Risk for Violence in Prison and Recidivism». En: Jay P. Singh, Daryl G. Kroner, J. Stephen Wormith, Sarah L. Desmarais, Zachary Hamilton (eds). Handbook of Recidivism Risk/Needs Assessment Tools, págs. 255-268. John Wiley & Sons Ltd. DOI: https://doi.org/10.1002/9781119184256.ch13

BLOCH-WEHBA, H. (2020). «Automation in Moderation». Cornell International Law Journal, vol. 53, págs. 42-96 [en línea]. Disponible en: https://community.lawschool.cornell.edu/wp-content/uploads/2021/03/Bloch-Wehba-final.pdf

CRESPO MIGUEL, M.; DOMÍNGUEZ CABRERA, B. (2020). «Perspectivas de las tecnologías de Chatbot y su aplicación a las entrevistas de evaluación del lenguaje». Pragmalingüística, vol. 2, págs. 100-113. DOI: https://doi.org/10.25267/Pragmalinguistica.2020.iextra2.06

CASTRO-TOLEDO, F. J.; MIRÓ-LLINARES, F.; AGUERRI, J. C. (2023). «Data-Driven Criminal Justice in the age of algorithms: epistemic challenges and practical implications». Crim Law Forum, vol. 34, pags. 295-316. DOI: https://doi.org/10.1007/s10609-023-09454-y

DIAS OLIVA, T.; ANTONIALLI, D. M.; GOMES, A. (2021). «Fighting Hate Speech, Silencing Drag Queens? Artificial Intelligence in Content Moderation and Risks to LGBTQ Voices Online». Sexuality & Culture, vol. 25, n.º 2, págs. 700-732. DOI: https://doi.org/10.1007/s12119-020-09790-w

DUARTE, N.; LLANSÓ, E. (2017). Mixed Messages? The Limits of Automated Social Media Content Analysis. CTD [en línea]. Disponible en: https://cdt.org/insights/mixed-messages-the-limits-of-automated-social-media-content-analysis/. [Fecha de consulta: 31 de mayo de 2023].

DUOEK, E. (2022). «Content Moderation as Systems Thinking». Harvard Law Review, vol. 136, 528-606. DOI: http://dx.doi.org/10.2139/ssrn.4005326

FEINBERG, J. (1985). Offense to Others (The Moral Limits of Criminal Law), vol. 2. Oxford: Oxford University press.

FUENTES OSORIO, J. L. (2021). «El odio como delito». Revista Electrónica de Ciencia Penal y Criminología, vol. 19-27, págs. 1-52 [en línea]. Disponible en: http://criminet.ugr.es/recpc/19/recpc19-27.pdf

GILLESPIE, T. (2020). «Content moderation, AI, and the question of scale». Big Data & Society, vol. 7, n.º 2. DOI: https://doi.org/10.1177/2053951720943234

GORWA, R.; BINNS, R.; KATZENBACH, C. (2020). «Algorithmic content moderation: Technical and political challenges in the automation of platform governance». Big Data & Society, vol. 7, n.º 1. DOI: https://doi.org/10.1177/2053951719897945

HELBERGER, N.; DIAKOPOULOS, N. (2023). «ChatGPT and the AI Act». Internet Policy Review, vol. 12, n.º 1. DOI: https://doi.org/10.14763/2023.1.1682

HUMAN RIGHTS WATCH (2020). «“Video Unavailable”. Social Media Platforms Remove Evidence of War Crimes». Human Rights Watch [en línea]. Disponible en: https://www.hrw.org/report/2020/09/10/video-unavailable/social-media-platforms-remove-evidence-war-crimes. [Fecha de consulta: 31 de mayo de 2023].

LLANSÓ, E. J. (2020). «No amount of “AI” in content moderation will solve filtering’s prior-restraint problem». Big Data & Society, vol. 7, n.º 1. DOI: https://doi.org/10.1177/2053951720920686

MARTÍNEZ-GARAY, L. (2018). “Peligrosidad, algoritmos y due process: El caso State vs. Loomis». Revista de Derecho Penal y Criminología, n.º 20, págs. 485-502. DOI: https://doi.org/10.5944/rdpc.20.2018.26484

MIRÓ LLINARES, F. (2015). «La criminalización de conductas “ofensivas”. A propósito del debate anglosajón sobre los “límites morales” del derecho penal». Revista electrónica de ciencia penal y criminología, n.º 17 [en línea]. Disponible en: http://criminet.ugr.es/recpc/17/recpc17-23.pdf

MIRÓ LLINARES, F. (2022). «Inteligencia artificial, delito y control penal: nuevas reflexiones y Algunas predicciones sobre su impacto en el derecho y la justicia penal». El Cronista del Estado Social y Democrático de Derecho, n.º 100, págs. 174-183.

MIRÓ-LLINARES, F.; MONEVA, A.; ESTEVE, M. (2018). «Hate is in the air! But where? Introducing an algorithm to detect hate speech in digital microenvironments». Crime Science, vol. 7, n.º 1. DOI: https://doi.org/10.1186/s40163-018-0089-1

OPENAI (2023a). «GPT-4 Technical Report». arXiv. DOI: https://doi.org/10.48550/arXiv.2303.08774

OPENAI (2023b). «Introducing ChatGPT». OpenAI Blog [en línea]. Disponible en: https://openai.com/blog/chatgpt#OpenAI. [Fecha de consulta: 21 abril 2023].

OUYANG, L.; WU, J.; JIANG, X.; ALMEIDA, D.; WAINWRIGHT, C. L.; MISHKIN, P.; ZHANG, C.; AGARWAL, S.; SLAMA, K.; RAY, A.; SCHULMAN, J.; HILTON, J.; KELTON, F.; MILLER, L.; SIMENS, M.; ASKELL, A.; WELINDER, P.; CHRISTIANO, P.; LEIKE, J.; LOWE, R. (2021). «Training language models to follow instructions with human feedback». arXiv. DOI: https://doi.org/10.48550/arXiv.2203.02155

PARIKH P. M.; SHAH D. M., PARIKH K. P. (2023). «Judge Juan Manuel Padilla Garcia, ChatGPT, and a controversial medicolegal milestone». Indian Journal of Medical Sciences, vol. 75, n.º 1, págs. 3-8. DOI: https://doi.org/10.25259/IJMS_31_2023

PETTINATO OLTZ, T. (2023, febrero). «ChatGPT, Professor of Law». SSRN. DOI: https://doi.org/10.2139/ssrn.4347630

PERLMAN, A. M. (2022). «The Implications of ChatGPT for Legal Services and Society». SSRN. DOI: https://doi.org/10.2139/ssrn.4294197

PRESNO LINERA, M. A. (2022). Derechos Fundamentales e Inteligencia Artificial. Madrid: Marcial Pons. DOI: https://doi.org/10.2307/jj.4908196

QUIJANO-SÁNCHEZ, L.; LIBERATORE, F.; CAMACHO-COLLADOS, J.; CAMACHO-COLLADOS, M. (2018). «Applying automatic text-based detection of deceptive language to police reports: Extracting behavioral patterns from a multi-step classification model to understand how we lie to the police». Knowledge-Based Systems, vol. 149, págs. 155-168. DOI: https://doi.org/10.1016/j.knosys.2018.03.010

RODRÍGUEZ MONTAÑÉS, T. (2012). Libertad de expresión, discurso extremo y delito. Una aproximación a las fronteras del derecho penal. Valencia: Tirant lo Blanch.

STOKEL-WALTER, C. (2023). «Generative AI Is Coming for the Lawyers. Large law firms are using a tool made by OpenAI to research and write legal documents. What could go wrong?». TheWired [en línea]. Disponible en: https://www.wired.co.uk/article/generative-ai-is-coming-for-the-lawyers. [Fecha de consulta: 31 de mayo de 2023].

TERUEL LOZANO, G. (2017). «El discurso del odio como límite a la libertad de expresión en el marco del convenio europeo». ReDCE, n.º 27 [en línea]. Disponible en: https://www.ugr.es/~redce/REDCE27/articulos/03_TERUEL.htm

UDUPA, S.; MARONIKOLAKIS, A.; SCHÜTZE, H.; WISIOREK, A. (2022). «Ethical Scaling for Content Moderation: Extreme Speech and the (In)Significance of Artificial Intelligence». Big Data & Society, vol. 10, n.º 1. DOI: https://doi.org/10.1177/20539517231172424. [Fecha de consulta: 31 de mayo de 2023]