As confusões de um conhecido e eficiente catedrático da universidade de Lisboa

Num artigo publicado no caderno de economia do semanário Expresso, o antigo presidente do IST, o catedrático Arlindo Oliveira, escreve que há uma obrigação moral para usar a IA por conta da eficiência que ela promove e vai ao ponto de afirmar que essa obrigação não é só das empresas mas também do Estado Português. Aquilo que ele porém se esqueceu foi de esclarecer qual o artigo ou artigos da Constituição da República Portuguesa (CRP) que permitem perceber essa obrigação moral. Ou talvez o referido catedrático deseje uma revisão da CRP para que a eficiência possa no futuro passar a ter o mesmo valor das obrigações morais vertidas na mesma, que incluem por exemplo, a dignidade da pessoa humana,  o direito à vida, à integridade pessoal, à liberdade e segurança etc etc. 

A eficiência é importante, mas esperemos que neste país não se chegue ao ponto (de com o apoio do Chega) ela seja elevada a uma tal categoria. De mais a mais, a adopção cega de IA, motivada por uma obsessão pelo aumento da eficiência, pode sim contribuir para violar várias obrigações morais vertidas na referida CRP, como seja por exemplo a automatização de serviços públicos, que pode excluir cidadãos idosos, ou com baixa ou mesmo nula literacia digital, pois recorde-se que Portugal possui ainda no século XXI, algumas centenas de milhares de pessoas analfabetas, ou como seja por exemplo pela utilização da IA na determinação de benefícios sociais, que correu de forma tão dramática na Holanda, quando mais de 26 000 famílias foram injustamente acusadas de fraude e que levou à queda do Governo e ao pagamento de mais de mil milhões de euros em indemnizações a essas familias  https://observador.pt/opiniao/o-algoritmo-que-derrubou-um-governo/

Pessoalmente e no que diz respeito à IA acho que o ex-presidente do IST, devia estar menos ocupado a produzir sentenças avulsas sobre moralidade (uma área em que já se percebeu não é especialista) e muito mais preocupado com o dilema que enfrenta o actual presidente do IST, e que noutros países já se reflecte na fuga dos estudantes do ensino superior, por conta da elevada probabilidade da sua formação se tornar redundante, muito antes ainda daqueles a terem terminado, vide artigo publicado há poucos dias na Forbes https://www.forbes.com/sites/victoriafeng/2025/08/06/fear-of-super-intelligent-ai-is-driving-harvard-and-mit-students-to-drop-out/

PS - Ou talvez o catedrático Arlindo Oliveira não tenha ainda alcançado os dilemas morais, que agora se levantam aos recém-diplomados, relativamente aos quais um artigo publicado na conhecida revista The Economist, afirmou que estão "lixados" (screwed) https://19-pacheco-torgal-19.blogspot.com/2025/06/a-tragedia-dos-milhoes-de-recem.html

My letter to the editor of the Journal of Informetrics criticising the paper "From 'Sleeping Beauties' to 'Rising Stars'

See below the text of my letter criticizing the thesis proposed by J. Gorraiz, the author of the paper, who is affiliated with the University of Vienna.

---

J. Gorraiz’s recently published paper presents a conceptually stimulating and metaphor-laden examination of the ideological foundations of bibliometrics, tracing their origins to religious, moral, and philosophical traditions. While such a reflective approach is thought-provoking, the paper suffers from several substantive limitations—particularly when read in light of ongoing debates surrounding the practical utility, cost-efficiency, and predictive power of bibliometric tools in research evaluation.

The paper’s central thesis—that bibliometrics derive from religious and philosophical traditions—is built on an extended metaphorical scaffolding. Citations are likened to divine judgment, H-indexes to spiritual tallies, and “sleeping beauties” to secular miracles. While these metaphors may have rhetorical appeal, they ultimately distract from more pressing empirical and methodological issues. There is no engagement with recent literature on field-normalized citation metrics, responsible metric frameworks (such as the Leiden Manifesto or DORA), or citation dynamics in different disciplines. Nor does the paper propose concrete methodological or policy alternatives. The result is a text rich in allegory but impoverished in evidence, leaving readers without clear guidance for improving bibliometric practice.
 
Gorraiz asserts that a low citation count does not imply irrelevance; it may reflect novelty. While this claim holds some truth, the argument is selectively framed and omits crucial empirical counterevidence. Notably, he fails to mention the robust findings from Clarivate Analytics, whose “Citation Laureates” methodology—based on identifying papers with exceptionally high citation counts (over 1,000)—has successfully predicted more than 70 Nobel Prize winners. Moreover, it is worth recalling the analysis by Traag and Waltman (2019), which demonstrated that citation-based metrics exhibit a strong correspondence with expert peer review assessments, particularly in fields such as Physics, Clinical Medicine, and Public Health. 

These findings collectively suggest that, far from being inherently unreliable, well-calibrated citation metrics can serve as a meaningful and practical complement—or, in some contexts, a viable alternative—to traditional peer review in the evaluation of research performance. These results strongly suggest that novelty and high citation impact are not mutually exclusive, and in fact, may often coincide. By disregarding this evidence, the paper constructs a false dichotomy between citation count and originality, while ignoring one of the most compelling demonstrations of bibliometrics' predictive capacity.
 
A still more consequential omission in the author’s analysis lies in the near-total absence of engagement with the underlying economic rationale for the widespread adoption of bibliometric tools. While the discussion frames citation indicators primarily as symbolic gestures or ritualistic artefacts within the academic system, it largely overlooks their pragmatic role as scalable and cost-efficient proxies in research evaluation—particularly in contexts where peer review faces severe logistical and financial constraints. Peer review, though indispensable in certain contexts, is notoriously resource-intensive: national research assessments such as the UK’s Research Excellence Framework (REF) have incurred costs exceeding £250 million per evaluation cycle. Similar pressures are evident in hiring processes, tenure reviews, and grant allocation panels, all of which require substantial investments of time, coordination, and expert labour. 

Crucially, empirical evidence undermines the dismissive treatment of bibliometrics: Abramo et al. (2019) demonstrated that citation-based indicators not only outperform peer review in predicting subsequent scholarly impact, but also exhibit increasing predictive accuracy over time. These findings bring into sharp relief the structural trade-offs between speed, cost, and precision that evaluation systems must navigate. Bibliometric measures—despite their well-known limitations—offer reproducible, transparent, and broadly applicable screening mechanisms capable of alleviating the evaluative burden on human reviewers. Any critique that ignores these economic and operational realities, while failing to articulate a credible alternative framework, risks producing an analysis that is philosophically stimulating yet practically inert in the policy and administrative domains where evaluation decisions are actually made.
 
Which approach is more detrimental to the progress of science: implementing a hybrid model of abbreviated peer review augmented by quantitative metrics—thereby conserving substantial financial resources—or relying exclusively on comprehensive, resource-intensive peer review protocols that allocate those funds away from direct research support? Moreover, how might the latter paradigm exacerbate inequities in research assessment for low-income countries, which lack the financial capacity to underwrite such costly evaluation processes?
 
Finally, allow me to provide you with some insights into my homeland, Portugal, which has experimented with both approaches. In a prior Portuguese research assessment conducted in 2013, the international experts serving on the evaluation panels enjoyed complete autonomy. They had the freedom to evaluate research units through on-site visits and also had access to a comprehensive bibliometric analysis, utilizing data from Scopus, which was expertly conducted by Elsevier and generated a range of valuable metrics (Publications per FTE, Citations per FTE, h-index, Field-Weighted Citation Impact, Top cited publications, National and International Collaborations).
 
However, in recent years, we experienced a shift in perspective, with a Science Minister who shared similar sentiments with those critical of bibliometrics. During the most recent research assessment in 2018, which involved the evaluation of 348 research units comprising nearly 20,000 researchers, the Evaluation Guide clearly dictated that absolutely no metric could be used by the panels (note that all panels were composed by international experts, 51 from UK, 21 from USA, 17 from Germany, 17 from France, 11 from The Netherlands, 8 from Finland, 8 from Ireland, 7 from Switzerland, 6 from Sweden, 5 from Norway and also from other countries).
 
Nonetheless, once the research assessment had concluded, I conducted an extensive search through all the reports across various scientific areas. What I discovered was that the reviewers assigned significant importance to the quantity of publications and the perceived “quality” of journals, even though such considerations were expressly prohibited by the Evaluation Guide. I found that “publications”, “quartiles” and even “impact factors” were mentioned in the assessment reports more than 500 times. Meaning that in the absence of any metric the international experts (somewhat ironically) decide to use the worst of them all. Such findings lend strong support to the observations of Morgan-Thomas et al. (2024), who noted that the historically robust association between journal rankings and expert evaluations persists unabated, despite institutional endorsements of the principles articulated in DORA. This enduring pattern underscores a profound tension between formal evaluative guidelines and the implicit heuristics that experts continue to apply in practice.

Europe Under Fire: The Wrath of Tyrants and Tech Titans Threatening Its Survival

Giuliano da Empoli, an Italian-Swiss political essayist and professor at Sciences Po Paris, gave an extensive interview today to a major Portuguese newspaper about his latest book, The Hour of the Predator: Encounters with the Autocrats and Tech Billionaires Taking Over the World. Since the article is behind a paywall, I thought it would be helpful to share a brief summary in the form of three key excerpts from the interview:

"To simplify, there are two major possibilities...either we manage to impose democratic rules and our way of life onto the digital dimension of life; or the absence of rules, the law of the strongest, and the jungle law of today’s digital world will impose themselves on our private lives, our democracies, and our public life. In the first case, we will continue to have a democratic future; in the other, we will cease to have a democratic future"

I believe this is a very humiliating moment for Europe and Europeans. European integration was once, to some extent, the model that others looked up to—but that is no longer the case. And we are powerless and, at least in part, unable to defend ourselves.

The one thing all predators have in common is attacking Europe. Every day. Trump, Putin, Musk, Zuckerberg... Which means we bother them. We’re an obstacle to their plans. I believe we should become an even greater one"

Giuliano da Empoli may be right to draw attention to Europe’s vulnerabilities when confronting autocrats and tech billionaires. Yet, he may also overlook the deeper significance of the European project as a remarkable civilizational leap. As Michio Kaku noted decades ago, "the European Union is the beginning of a Type I economy… these European countries, which have slaughtered each other ever since the ice melted 10,000 years ago... they have banded together, put aside their differences to create the European Union." This perspective suggests that, despite its vulnerabilities, the European project embodies a commitment to transnational cooperation and long-term resilience, serving as a testament to humanity’s capacity for reconciliation and collective purpose—an audacious step toward planetary maturity. https://pacheco-torgal.blogspot.com/2019/10/the-role-of-academia-towards-type-1.html

PS - Regarding da Empoli’s book, I’d like to emphasize that his thesis resonates with another work I mentioned earlier on this blog in early July. At that time, I argued that in this particular context, "it becomes all the more urgent—indeed, imperative—that academia, as the last bastion of critical inquiry and intellectual integrity, reaffirms its uncompromising commitment to truth as a non-negotiable principle" https://19-pacheco-torgal-19.blogspot.com/2025/07/book-world-builders-we-are-entering-new.html