Halucynacje chatbotów a prawda: główne nurty debaty i ich interpretacje

• Authors: Jan Kreft , Monika Boguszewicz-Kreft , Barbara Cyrek

Title variants

EN

Chatbot Hallucinations vs. Truth: Mainstream Debates and Their Interpretations

Abstracts

EN

Generative artificial intelligence (AI) systems are able to create content by applying machine learning to large amounts of training data. This new data may include text (e.g. Bard by Google, LLaMa by Meta or ChatGPT by OpenAI) and visuals (e.g. Stable Diffusion or DALL-E by OpenAI) and audio (e.g. VALL-E by Microsoft). The level of sophistication of this content may make it indistinguishable from human work. However, chatbots are characterized by the so-called hallucinations, which are a new type of disinformation. The aim of the article is to identify the main trends in the debate on the effects of the use of artificial intelligence, with particular emphasis on disinformation involving chatbots in the media environment.

PL

Generatywne systemy sztucznej inteligencji (SI) są w stanie tworzyć treści medialne poprzez zastosowanie uczenia maszynowego do dużych ilości danych szkoleniowych. Te nowe dane mogą obejmować tekst (np. Bard firmy Google, LLaMa firmy Meta lub ChatGPT firmy OpenAI) oraz elementy wizualne (np. Stable Diffusion lub DALL-E OpenAI) i dźwięk (np. VALL-E firmy Microsoft). Stopień zaawansowania tych treści może czynić je nieodróżnialnymi od twórczości człowieka. Chatboty cechują się jednak tzw. halucynacjami, które w istotnej części są nowym rodzajem dezinformacji. Celem podjętych badań jest identyfikacja głównych nurtów debaty poświęconej skutkom wykorzystania sztucznej inteligencji ze szczególnym uwzględnieniem dezinformacji z udziałem chatbotów w środowisku mediów. W badaniu przyjęto metodę badawczą systematycznego przeglądu literatury ograniczającą m.in. błąd selekcji. Interpretacja głównych nurtów debaty skłania do wniosku, że dezinformacja chatbotów w postaci ich halucynacji jest znacząca pod względem skali, jest optymalizowana i personalizowana oraz ma istotny potencjał erodowania zaufania społecznego.

Keywords

EN

chatbots; disinformation; artificial intelligence

PL

chatboty; dezinformacja; sztuczna inteligencja

• Publisher: Katolicki Uniwersytet Lubelski Jana Pawła II. Towarzystwo Naukowe KUL
• Journal: Roczniki Nauk Społecznych
• Year: 2024
• Volume: 52
• Issue: 1
• Pages: 165-185

Dates

published

2024

Contributors

author

Jan Kreft

• Gdańsk University of Technology

• https://orcid.org/0000000341298424

author

Monika Boguszewicz-Kreft

• WSB Merito Gdańsk

• https://orcid.org/0000000292947175

author

Barbara Cyrek

• DSW University of Lower Silesia Wroclaw

• https://orcid.org/0000000232706548

References

• Adlakha V., BehnamGhader P., Lu X. H., Meade N. i Reddy S. (2023), Evaluating correctness and faithfulness of instruction-following models for question answering, ArXiv Preprint. https://doi.org/10.48550/arXiv.2307.16877
• Alkaissi H. i McFarlane S.I. (2023), Artificial hallucinations in ChatGPT: implications in scientific writing, Cureus 15, nr 2, artykuł nr e35179. https://doi.org/10.7759/cureus.35179
• Bengio S., Vinyals O., Jaitly N. i Shazeer N. (2015), Scheduled sampling for sequence prediction with recurrent neural networks. Advances in neural information processing systems 28, s. 1-28.
• Boguszewicz-Kreft M. (2021), Marketing doświadczeń: jak poruszyć zmysły, zaangażować emocje, zdobyć lojalność klientów? Warszawa: CeDeWu.
• Capuano N., Fenza G., Loia V. i Nota F.D. (2023), Content-Based Fake News Detection with machine and deep learning: a systematic review, Neurocomputing 540, nr 14, s. 91-103. https://www.doi.org/10.1016/j.neucom.2023.02.005
• Chen M., Tworek J., Jun H., Yuan Q., Pinto H. P. D. O., Kaplan J., … i Zaremba W. (2021), Evaluating large language models trained on code, ArXiv Preprint. https://doi.org/ 10.48550/arXiv.2107.03374
• Cheng X., Zhang X., Cohen J. i Mou J. (2022), Human vs. AI: Understanding the impact of anthropomorphism on consumer response to chatbots from the perspective of trust and relationship norms, Information Processing & Management 59, nr 3, artykuł nr 102940. https://doi.org/ 10.1016/j.ipm.2022.102940
• Colleoni E. i Corsaro D. (2022), Critical issues in artificial intelligence algorithms and their implications for digital marketing, [w:] R. Belk. R. Llamas (red.), The Routledge companion to digital consumption, Hoboken: Taylor and Francis, s. 166-177.
• Curran S., Lansley S. i Bethell O. (2023), Hallucination is the last thing you need, ArXiv Preprint. https://doi.org/10.48550/arXiv.2306.11520
• DiResta R. (2020), AI-Generated Text Is the Scariest Deepfake of All, https://www.wired. com/story/ai-generated-text-is-the-scariest-deepfake-of-all/ [dostęp: 19.12.2023]
• Du Y. (2023), Cooperative multi-agent learning in a complex world: challenges and solutions, Proceedings of the AAAI Conference on Artificial Intelligence 37, nr 13, s. 15436. https://doi.org/10.1609/aaai.v37i13.26803
• Dziri N., Madotto A., Zaiane O. i Bose A.J. (2021), Neural path hunter: Reducing hallucination in dialogue systems via path grounding, ArXiv Preprint. https://doi.org/10.48550/arXiv. 2104.08455
• Emsley R. (2023), ChatGPT: these are not hallucinations – they’re fabrications and falsifications, Schizophrenia 9, artykuł nr 52. https://doi.org/10.1038/s41537-023-00379-4
• Evans O., Cotton-Barratt O., Finnveden L., Bales A., Balwit A., Wills P., Righetti L. i Saunders W. (2021), Truthful AI: Developing and governing AI that does not lie, ArXiv Preprint. https://doi.org/10.48550/arXiv.2110.06674
• Evans O., Cotton-Barratt O., Finnveden L., Bales A., BalwitA., Wills P., … i Saunders W. (2021), Truthful AI: Developing and governing AI that does not lie, ArXiv Preprint. https://doi.org/ 10.48550/arXiv.2110.06674
• Feak C.B. i Swales J. (2009), Telling a research story: Writing a literature review, Ann Arbor: University of Michigan Press.
• Figar V.N. (2023), Metaphorical framings in The New York Times online press reports about ChatGPT, Philologia Mediana 15, nr 15, s. 381-398. https://www.doi.org/10.46630/ phm.15.2023.27
• Floridi L. i Chiriatti M. (2020), GPT-3: Its nature, scope, limits, and consequences, Minds and Machines, 30, s. 681-694. https://www.doi.org/10.1007/s11023-020-09548-1
• Griffin L.D., Kleinberg B., Mozes M., Mai K.T., Vau M., Caldwell M. i Marvor-Parker A. (2023), Susceptibility to Influence of Large Language Models, ArXiv Preprint. https://doi.org/10.48550/ arXiv.2303.06074
• Grobler A. (2000), Uteoretyzowanie, relatywizm i prawda, Przegląd Filozoficzny – Nowa Seria, nr 2 (34), s. 37-45.
• Guzman A.L. i Lewis S.C. (2019), Artificial intelligence and communication: A Human–Machine Communication research agenda, New Media & Society 22, nr 1, s. 70-86. https://doi.org/10.1177/1461444819858691
• Heaven W.D. (2022), Why Meta’s latest large language model only survived three days online, https://www.technologyreview.com/2022/11/18/1063487/meta-large-language-model-ai-only-survived-three-days-gpt-3-science/ [dostęp: 19.12.2023]
• Huang D., Harasim S.A. i Leccia F. (2023), Understanding the emotional experience on consumer behaviors: A study on ChatGPT service (student thesis), Jönköping International Business School.
• Huang M.H. i Rust R.T. (2018), Artificial intelligence in service, Journal of service research 21, nr 2, s. 155-172. https://www.doi.org/10.1177/1094670517752459
• Huang M.H., Rust R. i Maksimovic V. (2019), The feeling economy: Managing in the next generation of artificial intelligence (AI), California Management Review 61, nr 4, s. 43-65. https://doi.org/10.1177/0008125619863436
• Ji Z., Lee N., Frieske R., Yu T., Su D., Xu Y., Ishii E., Bang Y.J., Madotto A. i Fung P. (2023), Survey of hallucination in natural language generation, ACM Computing Surveys 55, nr 12, s. 1-38. https://doi.org/10.1145/3571730
• Kasai J., Kasai Y., Sakaguchi K., Yamada Y. i Radev D. (2023), Evaluating gpt-4 and chatgpt on japanese medical licensing examinations, ArXiv Preprint, https://doi.org/10.48550/arXiv. 2303.18027
• Kasai J., Kasai Y., Sakaguchi K., Yamada Y. i Radev D. (2023), Evaluating gpt-4 and chatgpt on japanese medical licensing examinations, ArXiv Preprint. https://doi.org/10.48550/arXiv. 2303.18027
• Khanam Z., Alwasel B.N., Sirafi H. i Rashid M. (2021), Fake news detection using machine learning approaches. IOP conference series: materials science and engineering 1099, nr 1, 012040). https://www.doi.org/10.1088/1757-899X/1099/1/012040
• Kim J.H., Kim J., Park J., Kim C., Jhang, J. i King B. (2023), When ChatGPT Gives Incorrect Answers: The Impact of Inaccurate Information by Generative AI on Tourism Decision-Making, Journal of Travel Research, Online First. https://doi.org/10.1177/00472875231212996
• Kitchenham B. (2004), Procedures for performing systematic reviews, Keele University Technical Report 33, s. 1-26.
• Kreft J. (2017), Algorithm as demiurge: a complex myth of new media, [w:] Strategic imperatives and core competencies in the era of robotics and artificial intelligence, Hershey: IGI Global, s. 146-166.
• Kreft J. (2018), Władza algorytmów: u źródeł potęgi Google i Facebooka, Kraków: Wydawnictwo Uniwersytetu Jagiellońskiego.
• Kreft J. (2022), Władza platform. Za fasadą Google, Facebooka i Spotify, Kraków: Universitas.
• Kreft J. (2023), Władza misjonarzy. Zmierzch i świt świeckiej religii w Dolinie Krzemowej, Kraków: Universitas.
• Kreft J., Boguszewicz-Kreft M. i Fydrych M. (2023), (Lost) Pride and Prejudice. Journalistic Identity Negotiation Versus the Automation of Content, Journalism Practice, Online First, 1-24. https://www.doi.org/10.1080/17512786.2023.2289177
• Kreft J., Boguszewicz-Kreft, M. i Hliebova, D. (2023), Under the Fire of Disinformation. Attitudes Towards Fake News in the Ukrainian Frozen War, Journalism Practice, s. 1-21. https://www.doi.org/10.1080/17512786.2023.2168209
• Kreps S., McCain R.M. i Brundage M. (2022), All the news that’s fit to fabricate: AI-generated text as a tool of media misinformation, Journal of experimental political science 9, nr 1, s. 104-117. https://www.doi.org/10.1017/XPS.2020.37
• Kshetri N., Dwivedi Y.K., Davenport T.H. i Panteli N. (2023), Generative artificial intelligence in marketing: Applications, opportunities, challenges, and research agenda, International Journal of Information Management 75, nr 6, artykuł nr 102716. https://doi.org/10.1016/ j.ijinfomgt.2023.102716
• Li C., Bi B., Yan M., Wang W. i Huang S. (2021), Addressing semantic drift in generative question answering with auxiliary extraction, [w:] Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing 2: Short Papers [online], Association for Computational Linguistics, s. 942-947.
• Li Z. (2023), The dark side of chatgpt: Legal and ethical challenges from stochastic parrots and hallucination, ArXiv Preprint. https://doi.org/10.48550/arXiv.2304.14347
• Lin S., Hilton J. i Evans O. (2021), TruthfulQA: Measuring How Models Mimic Human Falsehoods, ArXiv Preprint, https://doi.org/10.48550/arXiv.2109.07958
• Lucidi P. B. i Nardi D. (2018), Companion robots: the hallucinatory danger of human-robot interactions, [w:] Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society, s. 17-22. https://www.doi.org/10.1145/3278721.3278741
• Luvembe A.M., Li W., Li S., Liu F. i Xu G. (2023), Dual emotion based fake news detection: A deep attention-weight update approach. Information Processing & Management 60, nr 4, artykuł nr 103354. https://www.doi.org/10.1016/j.ipm.2023.103354
• Marr B. (2023), Chatgpt: What are hallucinations and why are they a problem for ai systems, Bernard Marr, https://bernardmarr.com/chatgpt-what-are-hallucinations-and-why-are-they-a-problem-for-ai-systems/ [dostęp: 19.12.2023].
• Martinho A., Poulsen A., Kroesen M. i Chorus C. (2021), Perspectives about artificial moral agents, AI Ethics 1, s. 477-490. https://www.doi.org/10.1007/s43681-021-00055-2
• Mazur Z. i Orłowska A. (2018), Jak zaplanować i przeprowadzić systematyczny przegląd literatury, Polskie Forum Psychologiczne 23, nr 2, s. 235-251. https://doi.org/10.14656/PFP20180202
• Mbakwe A.B., Lourentzou I., Celi L.A., Mechanic O.J. i Dagan A. (2023), ChatGPT passing USMLE shines a spotlight on the flaws of medical education, PLOS Digital Health 2, nr 2, artykuł nr e0000205. https://doi.org/10.1371/journal.pdig.0000205
• McGuire J., De Cremer D., Hesselbarth Y., De Schutter L., Mai K. M. i Van Hiel A. (2023), The reputational and ethical consequences of deceptive chatbot use, Scientific Reports 13, artykuł nr 16246. https://www.doi.org/10.1038/s41598-023-41692-3
• McIntosh T.R., Liu T., Susnjak T., Watters P., Ng A. i Halgamuge M.N. (2023), A culturally sensitive test to evaluate nuanced gpt hallucination, IEEE Transactions on Artificial Intelligence. https://www.doi.org/10.1109/TAI.2023.3332837
• Metz C. (2023), Chatbots May ‘Hallucinate’ More Often Than Many Realize, The New York Times, https://www.nytimes.com/2023/11/06/technology/chatbots-hallucination-rates.html [dostęp: 19.12.2023].
• Milosavljević M. i Vobič I. (2019), Human still in the loop: Editors reconsider the ideals of professional journalism through automation, Digital Journalism 7, nr 8, s. 1098-1116. https://www.doi.org/10.1080/21670811.2019.1601576
• Mincewicz K. (2022), Sposoby pojmowania prawdy w prawoznawstwie na tle filozoficznych koncepcji prawdy. Szczecin: Uniwersytet Szczeciński.
• Morley J., Floridi L., Kinsey L. i Elhalal A. (2020), From what to how: An initial review of publicly available AI ethics tools, methods and research to translate principles into practices, Science and Engineering Ethics, 26, nr 4 s. 2141-2168. https://doi.org/10.1007/s11948-019-00165-5
• Mridha M.F., Keya A.J., Hamid, M.A., Monowar M.M. i Rahman M.S. (2021), A comprehensive review on fake news detection with deep learning, IEEE Access, 9, s. 156151-156170. https://www.doi.org/10.1109/ACCESS.2021.3129329
• Mukherjee A. i Chang H. (2023), The Creative Frontier of Generative AI: Managing the Novelty-Usefulness Tradeoff, ArXiv Preprint. https://doi.org/10.48550/arXiv.2306.03601
• Munn L., Magee L. i Arora V. (2023), Truth Machines: Synthesizing Veracity in AI Language Models, ArXiv Preprint. https://doi.org/10.48550/arXiv.2301.12066
• Murtarelli G., Gregory A. i Romenti S. (2021), A conversation-based perspective for shaping ethical human–machine interactions: The particular challenge of chatbots, Journal of Business Research 129, s. 927-935. https://doi.org/10.1016/j.jbusres.2020.09.018
• Oravec J.A., (2022), The emergence of “truth machines”?: Artificial intelligence approaches to lie detection, Ethics and Information Technology 24, nr 1, artykuł nr 6. https://www.doi.org/ 10.1007/s10676-022-09621-6
• Orr W. i Davis J.L. (2020), Attributions of ethical responsibility by artificial intelligence practitioners. Information, Communication & Society 23, nr 5, p. 719-735. https://www.doi.org/10.1080/ 1369118X.2020.1713842
• Pan Y., Froese F., Liu N., Hu Y. i Ye M. (2022), The adoption of artificial intelligence in employee recruitment: The influence of contextual factors, The International Journal of Human Resource Management 33, nr 6, s. 1125-1147. https://doi.org/10.1080/09585192.2021.1879206
• Paul J., Ueno A. i Dennis C. (2023), ChatGPT and consumers: Benefits, pitfalls and future research agenda, International Journal of Consumer Studies 47, nr 4, s. 1213-1225. https://doi.org/ 10.1111/ijcs.12928
• Pool J., Akhlaghpour S., Fatehi F. i Burton-Jones A. (2024), A systematic analysis of failures in protecting personal health data: a scoping review, International Journal of Information Management 74, artykuł nr 102719. https://doi.org/10.1016/j.ijinfomgt.2023.102719
• Rawte V., Chakraborty S., Pathak A., Sarkar A., Tonmoy S. M., Chadha A., … i Das A. (2023), The Troubling Emergence of Hallucination in Large Language Models--An Extensive Definition, Quantification, and Prescriptive Remediations, ArXiv Preprint. https://doi.org/10.48550/ arXiv.2310.04988
• Ricaurte P. (2022), Ethics for the majority world: AI and the question of violence at scale. Media, Culture & Society 44, nr 4, s. 726-745. https://www.doi.org/10.1177/01634437221099612
• Rosen J. (1993), Beyond Objectivity, Nieman Reports 47, nr 4, s. 48-53.
• Runco M.A. (2023), AI can only produce artificial creativity, Journal of Creativity 33, nr 3, artykuł nr 100063. https://doi.org/10.1016/j.yjoc.2023.100063
• Schull N. (2013), The Folly of technological Soultionism: an Interview with Evgeny Morozov, Public Books, https://www.publicbooks.org/the-folly-of-technological-solutionism-an-interview-with-evgeny-morozov/ [dostęp: 19.12.2023].
• Shin D. (2021), The effects of explainability and causability on perception, trust, and acceptance: Implications for explainable AI, International Journal of Human-Computer Studies 146, artykuł nr 102551. https://www.doi.org/10.1016/j.ijhcs.2020.102551
• Shu K., Sliva A., Wang S., Tang J. i Liu H. (2017), Fake news detection on social media: A data mining perspective, ACM SIGKDD explorations newsletter 19, nr 1, s. 22-36. https://www.doi.org/10.1145/3137597.3137600
• Shu K., Wang S. i Liu H. (2019), Beyond news contents: The role of social context for fake news detection, [w:] Proceedings of the twelfth ACM international conference on web search and data mining, s. 312-320. https://doi.org/10.48550/arXiv.1712.07709
• Sidaoui K., Jaakkola M. i Burton J. (2020), AI feel you: customer experience assessment via chatbot interviews, Journal of Service Management 31, nr 4, s. 745-766. https://www.doi.org/ 10.1108/JOSM-11-2019-0341
• Solaiman I. i Dennison C. (2021), Process for Adapting Language Models to Society (PALMS) with Values-Targeted Datasets, ArXiv Preprint. https://doi.org/10.48550/arXiv.2106.10328
• Stachewicz K. (2013), O filozofii chrześcijańskiej Kilka uwag z perspektywy historycznej i futurologicznej, Logos i Ethos, nr 2(35), s. 219-234.
• Sun K., Xu Y.E., Zha H., Liu Y. i Dong X.L. (2023), Head-to-Tail: How Knowledgeable are Large Language Models (LLM)? AKA Will LLMs Replace Knowledge Graphs?, ArXiv Preprint. https://doi.org/10.48550/arXiv.2308.10168
• Swart J., Groot Kormelink T., Costera Meijer I. i Broersma M. (2022), Advancing a radical audience turn in journalism. Fundamental dilemmas for journalism studies, Digital Journalism 10, nr 1, s. 8-22. https://doi.org/10.1080/21670811.2021.2024764
• Tandoc Jr E.C. i Lee J.C.B. (2022), When viruses and misinformation spread: How young Singaporeans navigated uncertainty in the early stages of the COVID-19 outbreak, New Media & Society 24, nr 3, s. 778-796. https://doi.org/10.1177/1461444820968212
• Touvron H., Martin L., Stone K., Albert P., Almahairi A., Babaei Y., … i Scialom T., (2023), Llama 2: Open foundation and fine-tuned chat models, ArXiv Preprint. https://doi.org/10.48550/arXiv.2307.09288
• Tranfield D., Denye D. i Smart P. (2003), Towards a Methodology for Developing Evidence-Informed Management Knowledge by Means of Systematic Review, British Journal of Management 14, nr 3, s. 207-222. https://www.doi.org/10.1111/1467-8551.00375
• Umapathi L.K., Pal A. i Sankarasubbu M. (2023), Med-halt: Medical domain hallucination test for large language models, ArXiv Preprint. https://doi.org/10.48550/arXiv.2307.15343
• Verme P., Oremus W. (2023), What GPT invented a sexual harassment scandal and named a real law prof as the accused, https://www.washingtonpost.com/technology/2023/04/05/chatgpt-lies/ [dostęp: 19.12.2023].
• Waisbord S. (2018), Truth is what happens to news: On journalism, fake news, and post-truth, Journalism Studies 19, nr 13, s. 1866-1878. https://doi.org/10.1080/1461670X.2018.1492881
• Wamba S.F., Queiroz M.M., Jabbour C.J.C. i Shi C.V. (2023), Are both generative AI and ChatGPT game changers for 21st-Century operations and supply chain excellence?, International Journal of Production Economics 265, artykuł nr 109015. https://www.doi.org/10.1016/ j.ijpe.2023.109015
• Wang H., Fu T., Du Y., Gao W., Huang K., Liu Z., … i Zitnik M. (2023), Scientific discovery in the age of artificial intelligence, Nature, 620, s. 47-60. https://www.doi.org/10.1038/s41586-023-06221-2
• Wiseman S., Shieber S.M., i Rush A.M. (2017), Challenges in data-to-document generation, ArXiv Preprint. https://doi.org/10.48550/arXiv.1707.08052
• Woleński J. (2013), Historia pojęcia prawdy, [w:] R. Ziemińska (red.), Przewodnik po epistemologii, Kraków: WAM, s. 53-86.
• Zarouali B., Makhortykh M., Bastian M. i Araujo T. (2021), Overcoming polarization with chatbot news? Investigating the impact of news content containing opposing views on agreement and credibility, European Journal of Communication 36, nr 1, s. 53-68. https://doi.org/10.1177/ 0267323120940908

Identifiers

DOI

10.18290/rns2024.0009

Biblioteka Nauki

30147145