Acessibilidade / Reportar erro

ChatGPT: a museum of great novelties

In recent years, the field of Artificial Intelligence (AI) has advanced considerably. More recently, the great object of discussion in academia has been the ChatGPT (Marcus, Davis, & Aaronson, 2022Marcus, G., Davis, E., & Aaronson, S. (2022, maio 02). A very preliminary analysis of DALL-E 2. Recuperado de https://doi.org/10.48550/arXiv.2204.13807
https://doi.org/10.48550/arXiv.2204.1380...
; Rossoni & ChatGPT, 2023Rossoni, L., & ChatGPT. (2022). A inteligência artificial e eu: escrevendo o editorial juntamente com o ChatGPT. Revista Eletrônica de Ciência Administrativa, 21(3), 399-405. Recuperado dehttps://doi.org/10.21529/RECADM.2022ed3
https://doi.org/10.21529/RECADM.2022ed3...
).

This public tool developed by OpenAI (Brockman et al., 2016Brockman, G., Cheung, V., Pettersson, L., Schneider, J., Schulman, J., Tang, J., … Zaremba, W. (2016, junho 05). OpenAI Gym. Recuperado de https://doi.org/10.48550/arXiv.1606.01540
https://doi.org/10.48550/arXiv.1606.0154...
), based on the Generative Pre-Trained Transformer (GPT) language model, is a chatbot capable of meeting a range of requests from its users (Kirmani, 2022Kirmani, A. R. (2022). Artificial intelligence-enabled science poetry. ACS Energy Letters, 8(1), 574-576. Recuperado de https://doi.org/10.1021/acsenergylett.2c02758
https://doi.org/10.1021/acsenergylett.2c...
).

As laypeople, we might ask ourselves: what is a chatbot? What is GPT?

By definition, a chatbot is a computer program designed to simulate conversations with human users, especially over the Internet (King, 2023King, M. R. (2023). The future of AI in medicine: A perspective from a chatbot. Annals of Biomedical Engineering, 51, 291-295. Recuperado de https://doi.org/10.1007/s10439-022-03121-w
https://doi.org/10.1007/s10439-022-03121...
). GPT is an artificial learning model, which uses unsupervised and supervised learning techniques to understand and generate human-like language (Radford, Narasimhan, Salimans, & Sutskever, 2018Radford, A., Narasimhan, K., Salimans, T., & Sutskever, I. (2018). Improving language understanding by generative pre-training. Recuperado dehttps://www.cs.ubc.ca/~amuham01/LING530/papers/radford2018improving.pdf
https://www.cs.ubc.ca/~amuham01/LING530/...
).

ChatGPT can answer simple questions and write more complex texts (Liu et al., 2021Liu, X., Zheng, Y., Du, Z., Ding, M., Qian, Y., Yang, Z., … Tang, J. (2021, março 18). GPT understands, too. Recuperado de https:// doi.org/10.48550/arXiv.2103.10385
https://doi.org/10.48550/arXiv.2103.1038...
) in a language almost indistinguishable from natural human language (Dale, 2021Dale, R. (2021). GPT-3 What’s it good for? Natural Language Engineering, 27(1), 113-118. Recuperado de https://doi.org/10.1017/S1351324920000601
https://doi.org/10.1017/S135132492000060...
). This is one of the factors that justifies its rapid popularization (Mollman, 2022Mollman, S. (2022, dezembro 09). ChatGPT gained 1 million users in under a week. Here’s why the AI chatbot is primed to disrupt search as we know it. Fortune. Recuperado dehttps://fortune.com/2022/12/09/ai-chatbot-chatgpt-could-disrupt-google-search-engines-business/
https://fortune.com/2022/12/09/ai-chatbo...
).

However, ChatGPT has its limitations. Since it is built on a statistical basis using standards defined by a large set of text data, there is a possibility that prejudices and stereotypes present in the data are replicated (Dale, 2017Dale, R. (2017). NLP in a post-truth world. Natural Language Engineering, 23(2), 319-324. Recuperado de https://doi.org/10.1017/S1351324917000018
https://doi.org/10.1017/S135132491700001...
; Lucy & Bamman, 2021Lucy, L., & Bamman, D. (2021). Gender and representation bias in GPT-3 generated stories. In Proceedings of the 3º Workshop on Narrative Understanding, online.), meaning the final essay may contain offensive or discriminatory phrases.

Furthermore, GPT models cannot fully understand the context and meaning of the text (Strubell, Ganesh, & McCallum, 2019Strubell, E., Ganesh, A., & McCallum, A. (2019). Energy and policy considerations for deep learning in NLP. In Proceedings of the 57º Annual Meeting of the Association for Computational Linguistics, Florence, Italy.), given that these are invariably socially constructed (Berger & Luckmann, 2014Berger, P. L., & Luckmann, T. (2014). A construção social da realidade: tratado de sociologia do conhecimento. Petrópolis, RJ: Vozes.). Another point to be considered is the operational costs: energy to process these algorithms and store the generated data (Zhou, Chen, Jin, & Wang, 2021Zhou, X., Chen, Z., Jin, X., & Wang, W. Y. (2021). HULK: An energy efficiency benchmark platform for responsible natural language processing. In Proceedings of the16ºConference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, online.) and computational resources (Lund & Wang, 2023Lund, B. D., & Wang, T. (2023, janeiro 22). Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News. Recuperado de http://dx.doi.org/10.2139/ssrn.4333415
https://doi.org/10.2139/ssrn.4333415...
).

So, what opportunities and challenges does ChatGPT offer to professors, researchers, reviewers, and editors?

On the one hand, this tool optimizes time and effort in producing, revising, and editing texts, as it can efficiently format and organize the content. On the other hand, Lund and Wang (2023Lund, B. D., & Wang, T. (2023, janeiro 22). Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News. Recuperado de http://dx.doi.org/10.2139/ssrn.4333415
https://doi.org/10.2139/ssrn.4333415...
) argue that these tools may compromise the scientific integrity of articles to the extent that, although AI algorithms are designed to be objective, they can still be influenced by data used to program them or by the prejudices of the humans who design them.

As researchers, reviewers, and editors, we want to deepen this discussion. A big mistake is to take the theoretical framework as an extensive sequence of citations, the so-called “name-dropping.” Instead, the authors must use this section to revisit the literature and actually discuss the chosen authors, their ontological perspectives, and to what extent their points of view are congruent or divergent.

Likewise, in empirical research, the presentation of the methodological path is totally authorial and reflective; therefore, it cannot be artificially produced.

Finally, the same can be said about the research findings, the discussion of results, and recommendations proposing research agendas. These sections demand reflection and the researcher’s immersion in the object of study.

Let us do a mental exercise. Imagine that, in a call for papers, five authors used ChatGPT to produce an article on the same topic. There is a great possibility that all texts would be very similar. Thus, the anti-plagiarism system would identify the lack of copyright, and the articles would be rejected.

We advocate that the misuse of AI tools - for instance, manipulating or distorting scientific records - should be strictly penalized. Unfortunately, in the United States, there have already been cases of researchers using AI tools to generate false documents or manipulate the results of experiments (Lund & Wang, 2023Lund, B. D., & Wang, T. (2023, janeiro 22). Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News. Recuperado de http://dx.doi.org/10.2139/ssrn.4333415
https://doi.org/10.2139/ssrn.4333415...
).

This discussion should also be taken to our classrooms since our mission is to train professionals capable of thinking critically, identifying opportunities and challenges for organizations, society, and governments, and devising solutions to complex problems. In that case, simply typing a question into a computer will not make them develop such skills.

Ultimately, for academia, AI is “a museum of great novelties,” i.e., it brings back the old and well-known concerns regarding plagiarism (Irigaray, 2020Irigaray, H. A. R. (2020). Plágio e pirataria na academia: entre Mizner e o Código Penal Brasileiro. Cadernos EBAPE.BR, 18(3), 1-6. Recuperado de https://doi.org/10.1590/1679-395181801
https://doi.org/10.1590/1679-395181801...
) while presenting new and exciting paths with challenges and traps for the near future.

While we are on the brink of this future, let us look at the past...

As we do every first editorial of the year, we want to present the consolidated data of Cadernos EBAPE.BR’s previous year.

In 2022, the journal received 318 submissions, 46 papers written in English and five in Spanish. Of these, 166 were accepted (52%), 142 were rejected (45%), and 10 were withdraw (3%).

Last year, we published 66 articles in six issues. They were written by 165 authors affiliated with national institutions and 08 authors from seven international institutions.

One of the biggest challenges for us editors is to reduce the average time between submission, desk review, final editorial decision, and publication. As we work with a continuous desk review, authors receive the first response within a week of submission. Thanks to our team of evaluators, the average time between submission and the final editorial decision has been reduced to three months and between submission and publication to ten months.

We would also like to emphasize our engagement in disseminating the Brazilian academic production, offering the opportunity for a fast-track review of some studies presented at the EnANPAD, EnANGRAD, and ENGEMA meetings in the life of people management and labor relations. We reiterate our willingness to establish other partnerships.

In 2022, there were five calls for papers. The first, entitled “Infrastructure delivery and project management in developing and emerging economies,” received 17 abstracts and eight manuscripts, four of which were accepted for publication.

In the second, “Debating black slavery in Management and Organization Studies from decolonial and Afro-Diasporic perspectives,” we received 18 manuscripts, of which seven were accepted for publication, and two are still under review.

In the third call for papers, “Labor, migration, and mobility: a Lusophone dialogue,” we received 13 manuscripts, of which three are still under review.

In the fourth, “Leadership: revisiting and reframing the big questions of theory and practice,” 27 manuscripts were received, of which seven are still under review.

Finally, in the fifth, “Critical thinking vs. organizational thinking,” we received 29 manuscripts, 18 of which are still under review.

For 2023, we have already released two calls for papers: “Social and solidarity economy in the organization of decent work: sociological interpretations,” with a deadline of January 31, and “Decolonizing perspectives and decolonial pluriversality in management praxis & research,” with a deadline of March 15.

Cadernos EBAPE.BR maintained its classification as an A2 journal for the next quadrennium. In addition, 2023 is a very special year for us since the journal completes its 20th anniversary and will begin adopting continuous publication, Open Science, and the Contributor Roles Taxonomy (CREdiT) specification system.

This first issue of 2023 starts with the article “Fake news and storytelling: two sides of the same coin or two identical coins?” It is a theoretical essay that proposes that both narratives that make up fake news and storytelling are equal, as they have similarities in the processes of making, reproducing, and, mainly, in their ulterior motive - to maintain or obtain economic, social, or political capital.

In “The role of race relations in the Brazilian labor market: recruitment and selection processes in focus,” Cláudia Aparecida Avelar Ferreira, Simone Costa Nunes, and Jair Nascimento Santos analyze how racial relations present in Brazilian society intersect gender and social class and influence the inclusion of Black and White women in the formal labor market, through the speeches of Brazilian students and recruiters who work for their own companies or for others, national and international.

In the third article, “The Relationship between self-efficacy and organizational reputation in cooperative organizations,” Vanessa Christofoli and Alex Sandro Quadros Weymer seek to identify the relationship between self-efficacy and organizational reputation, based on the contribution of graduates of a professional master’s degree in their respective cooperatives.

In “Flexible work as a rule in capitalism: conceptualization and theoretical-analytical propositions,” Geraldo Tessarini Junior, Patrícia Saltorato, and Kaio Lucas da Silva Rosa explore the theme of flexibilization of work in the context of flexible capitalism to propose a theoretical-analytical analysis model related to the different classifications of this phenomenon.

Rosana da Rosa Portella Tondolo, Vilmar Antonio Gonçalves Tondolo, Cláudia Cristina Bitencourt, and Ely Laureano Paiva analyze the effect of transparency and the moderating effect of social capital in the intention of the managers of civil society organizations regarding fundraising in their article “Effect of transparency and social capital on fundraising intentions.”

Then, in “Institutionalizing markets: a proposed research agenda,” Francisco Cláudio Freitas Silva, Sérgio Fernando Loureiro Rezende, and Ramon Silva Leite promote a dialogue between institutional theory and marketing. They present an ascending research theme: the institutionalization of markets and propose a research agenda with questions related to the pillars of legitimacy, institutional logic, contested markets, and the spatiality of markets.

In the seventh article, entitled “Control and surveillance in digital capitalism: an analysis of blockchain technologies and their business implementation,” Pablo Emanuel Romero Almada and Elizardo Scarpati Costa use a case study to discuss how New Information and Communication Technologies (NICTs) interrelate with the systems’ control and surveillance.

Next, Gabriele Girardi elaborates a theoretical essay, “Dynamic capabilities based on knowledge and transformation in business models in the industry 4.0 scenario,” in which the author explores business models theories and dynamic capabilities based on knowledge in the scenario of digital transformation.

In “Socio-clinical analysis of the work context and its relationship with the mental illness of military police officers in the Federal District,” Cledinaldo Aparecido Dias, Marcus Vinicius Soares Siqueira, and Leonardo Borges Ferreira use clinical sociology and critical discourse analysis to investigate the work context of the state military police of the Federal District in Brazil and its impact on police officers’ mental health. This investigation was carried out through an ethnographic study.

The tenth article, “Autism in organizations: perceptions and actions for inclusion from the point of view of managers,” by Ana Teresa Oliveira da Silva Basto and Vanessa Martines Cepellos, aims to identify the perceptions and actions for inclusion, from the point of view of managers, of people with Autistic Spectrum Disorder (ASD) in the organizations that employ these professionals in Brazil.

Who are we and who are they? Historical transformations of violence in the human-animal relationship represented in artistic expressions,” by Renata Frota and Leticia Moreira Casotti, looks at transformations of violence present in the human-animal relationship in multiple periods through artistic expressions.

The case study “Trinks.com - digital platform at the service of beauty,” by Bruno Fernandes and Victor Manoel Cunha de Almeida, reports the dilemma faced by Marcel Gewerc, co-founder and CEO of Trinks.com. This company offered a digital solution to facilitate the internal management of beauty salons and allow customers to make appointments online.

We wish you a pleasant read.

Ph.D. Hélio Arthur Reis Irigaray

Editor-in-chief

Ph.D. Fabricio Stocker

Associate Editor

REFERÊNCIAS

  • Berger, P. L., & Luckmann, T. (2014). A construção social da realidade: tratado de sociologia do conhecimento Petrópolis, RJ: Vozes.
  • Brockman, G., Cheung, V., Pettersson, L., Schneider, J., Schulman, J., Tang, J., … Zaremba, W. (2016, junho 05). OpenAI Gym Recuperado de https://doi.org/10.48550/arXiv.1606.01540
    » https://doi.org/10.48550/arXiv.1606.01540
  • Dale, R. (2017). NLP in a post-truth world. Natural Language Engineering, 23(2), 319-324. Recuperado de https://doi.org/10.1017/S1351324917000018
    » https://doi.org/10.1017/S1351324917000018
  • Dale, R. (2021). GPT-3 What’s it good for? Natural Language Engineering, 27(1), 113-118. Recuperado de https://doi.org/10.1017/S1351324920000601
    » https://doi.org/10.1017/S1351324920000601
  • Irigaray, H. A. R. (2020). Plágio e pirataria na academia: entre Mizner e o Código Penal Brasileiro. Cadernos EBAPE.BR, 18(3), 1-6. Recuperado de https://doi.org/10.1590/1679-395181801
    » https://doi.org/10.1590/1679-395181801
  • King, M. R. (2023). The future of AI in medicine: A perspective from a chatbot. Annals of Biomedical Engineering, 51, 291-295. Recuperado de https://doi.org/10.1007/s10439-022-03121-w
    » https://doi.org/10.1007/s10439-022-03121-w
  • Kirmani, A. R. (2022). Artificial intelligence-enabled science poetry. ACS Energy Letters, 8(1), 574-576. Recuperado de https://doi.org/10.1021/acsenergylett.2c02758
    » https://doi.org/10.1021/acsenergylett.2c02758
  • Liu, X., Zheng, Y., Du, Z., Ding, M., Qian, Y., Yang, Z., … Tang, J. (2021, março 18). GPT understands, too Recuperado de https:// doi.org/10.48550/arXiv.2103.10385
    » https://doi.org/10.48550/arXiv.2103.10385
  • Lucy, L., & Bamman, D. (2021). Gender and representation bias in GPT-3 generated stories. In Proceedings of the Workshop on Narrative Understanding, online.
  • Lund, B. D., & Wang, T. (2023, janeiro 22). Chatting about ChatGPT: How may AI and GPT impact academia and libraries? Library Hi Tech News Recuperado de http://dx.doi.org/10.2139/ssrn.4333415
    » https://doi.org/10.2139/ssrn.4333415
  • Marcus, G., Davis, E., & Aaronson, S. (2022, maio 02). A very preliminary analysis of DALL-E 2 Recuperado de https://doi.org/10.48550/arXiv.2204.13807
    » https://doi.org/10.48550/arXiv.2204.13807
  • Mollman, S. (2022, dezembro 09). ChatGPT gained 1 million users in under a week. Here’s why the AI chatbot is primed to disrupt search as we know it. Fortune Recuperado dehttps://fortune.com/2022/12/09/ai-chatbot-chatgpt-could-disrupt-google-search-engines-business/
    » https://fortune.com/2022/12/09/ai-chatbot-chatgpt-could-disrupt-google-search-engines-business/
  • Radford, A., Narasimhan, K., Salimans, T., & Sutskever, I. (2018). Improving language understanding by generative pre-training Recuperado dehttps://www.cs.ubc.ca/~amuham01/LING530/papers/radford2018improving.pdf
    » https://www.cs.ubc.ca/~amuham01/LING530/papers/radford2018improving.pdf
  • Rossoni, L., & ChatGPT. (2022). A inteligência artificial e eu: escrevendo o editorial juntamente com o ChatGPT. Revista Eletrônica de Ciência Administrativa, 21(3), 399-405. Recuperado dehttps://doi.org/10.21529/RECADM.2022ed3
    » https://doi.org/10.21529/RECADM.2022ed3
  • Strubell, E., Ganesh, A., & McCallum, A. (2019). Energy and policy considerations for deep learning in NLP. In Proceedings of the 57º Annual Meeting of the Association for Computational Linguistics, Florence, Italy.
  • Zhou, X., Chen, Z., Jin, X., & Wang, W. Y. (2021). HULK: An energy efficiency benchmark platform for responsible natural language processing. In Proceedings of the16ºConference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, online.

Publication Dates

  • Publication in this collection
    13 Mar 2023
  • Date of issue
    Jan-Feb 2023
Fundação Getulio Vargas, Escola Brasileira de Administração Pública e de Empresas Rua Jornalista Orlando Dantas, 30 - sala 107, 22231-010 Rio de Janeiro/RJ Brasil, Tel.: (21) 3083-2731 - Rio de Janeiro - RJ - Brazil
E-mail: cadernosebape@fgv.br