Artificial intelligence, as a powerful personal assistant, is transforming our relationship with intellectual production, expression and the dissemination of ideas. It does not merely reduce the friction inherent in creation – whether writing, analysing or explaining – it also reshuffles the deck of influence and intellectual legitimacy.
AI, the end of the reign of rhetoric and an accelerator of ideas
For a long time, mastery of language, style and conceptual tools allowed some people to monopolise attention and exercise an authority that was sometimes undeserved. AI is challenging this dynamic by making highly demanding writing and analytical skills accessible to everyone. From now on, the value of an argument no longer depends on rhetorical skill or the technicality of the vocabulary used, but on the strength of the ideas it defends.
From this perspective, AI acts as a catalyst for clarity and discernment. Where the Sokal affair [1] exposed the manipulative use of jargon to lend an artificial aura of intellectual depth, AI, by democratising access to tools for structuring and building arguments, unmasks bluster. It allows collective attention to be focused on the relevance of content, hastening the downfall of weak ideas and propelling genuinely transformative proposals.
In the context of the transitions we must carry out – ecological, social and economic – this cognitive revolution opens up unprecedented prospects. It gives everyone back the ability to take an active part in debates and to propose solutions without being hindered by barriers of language or rhetoric. Ultimately, as strategy professor Philippe Silberzahn points out, might AI not represent a chance for the least qualified [2]?
Between irrational fears and transformative opportunities
More generally, it is essential not to fall into the trap of negativity bias [3], which consists in focusing more on the risks and excesses of AI (breaches of personal data, manipulation through deep fakes, feared job losses) than on its immense opportunities. As Philippe Silberzahn also stresses, automation has always prompted fears [4], but historically it has led to higher productivity, expanding markets and the creation of new jobs. Likewise, AI is not meant to replace humans, but to be a powerful tool that multiplies our capacities. Rather than yielding to fear or wait-and-see attitudes, we must learn to integrate it intelligently in order to maximise its transformative potential.
Another aspect often raised is the energy impact of AI [5]. According to the International Energy Agency, data centres currently account for around 1 to 2% of global electricity demand. While the rise of AI is increasing this consumption, it remains contained thanks to technological efficiency gains. Since 2008, the energy intensity of computer chips has fallen by more than 99%, making it possible to absorb the growth in usage without an explosion in energy consumption. According to a recent Goldman Sachs Research report [6] examining the impact of the rise of artificial intelligence, the share of global electricity consumption attributed to data centres could reach 3–4% by 2030, while their CO₂ emissions are expected to more than double, reaching 0.6% of global energy-related emissions. That said, although AI models consume more than a conventional internet search, their overall impact remains lower than that of other sectors such as industry, transport or air conditioning, which, for example, accounts for 10% of global electricity demand [7]. The energy question must therefore be approached with nuance, taking into account the optimisation efforts and the investments that technology companies are making in low-carbon energy.
Towards enlightened regulation, not paralysing regulation
Finally, it is essential to adopt a balanced approach to AI regulation. To be reasonable, regulation should be counterfactual, weighing the pros and cons rather than hiding behind a dogmatic precautionary principle. Any sensible regulation must rest on a trade-off between the risks of use and the risks of non-use [8]. Excessive caution can hold back major advances, notably in fields such as medicine, where AI makes it possible to speed up research and improve patient care. As the philosopher of science Maarten Boudry points out, precaution sometimes means taking action [9], and the adage that one should not act if there is a risk is simply unrealistic because, as he puts it: ‘Sometimes jumping is safer than standing still.’ This does not mean we should ignore that AI can cause harm, for example if a self-driving car were responsible for an accident. While regulation must ensure ethical and secure use, it must not stifle innovation by imposing disproportionate constraints. Comparing AI with an ideal, unrealistic situation prevents us from gauging its real impact relative to human alternatives, which are often imperfect. A pragmatic and adaptive approach is therefore needed to maximise its benefits while keeping its risks under control.
While the debate on the use of AI occupies editorial boards [10] in the press, many journalists are already using it daily, often discreetly and without admitting it publicly. For fear of criticism or internal disapproval, some young journalists prefer not to reveal that they rely on ChatGPT or other AI tools to structure their articles, suggest new angles, rephrase passages or even write entire pieces. As a result, confusion persists over the ethical and professional framework surrounding these practices: while some newsrooms strictly regulate the use of AI, others officially ban it while implicitly tolerating its use behind the scenes. This paradox reveals a transition under way in journalism, in which AI, rather than being seen as a threat, could come to be recognised as a legitimate professional asset.
A lever for critical thinking and emancipation
Along these lines, as early as April 2023, in his article ‘IA, la nouvelle machine à écrire’ [11], Antoine Buéno argued that, just as the arrival of the machine freed mathematicians from the drudgery of certain calculations, AI frees journalists from repetitive writing tasks, allowing them to refocus their profession on analysis and creativity. Like the calculator, which did not kill off mathematicians but allowed them to go further, ChatGPT is a writing aid rather than the gravedigger of journalism and literature. Although it poses a challenge to writers, who are taken aback by its creativity [12], it does not replace their soul. Likewise, although deepfakes are worrying, they could strengthen our critical faculties when faced with images and accelerate an anthropological shift in our relationship with reality [13]. Just as the printing press forced people to distinguish truth from falsehood by teaching them that not everything written is true, the proliferation of falsified images could make us more vigilant. Far from heralding a collapse of thought, AI could on the contrary be a lever for intellectual emancipation, obliging us to redefine our standards of truth and discernment. That said, this does not answer every question raised by the use of AI, notably the legitimate question of moral injury raised by many artists [14], for example.
In envisaging AI as a tool of intellectual emancipation whose consequences depend on how it is used, Antoine Buéno was more right than he knew. Contrary to the widespread idea that conspiracy theorists are gullible, an article in Psyche magazine [15] highlights their active engagement in seeking out information and building sophisticated arguments. The quest for discovery and the intellectual pleasure we derive from our own research play a key role in our adherence to various theories. In this respect, a 2024 preprint study [16] showed that interactions with ChatGPT-4 Turbo significantly and durably changed the beliefs of people who subscribed to certain theories. Rather than simply setting facts against received ideas, the model engaged in reasoning with users, questioning their assumptions and stimulating their critical thinking. This dialogue enabled a gradual repositioning of beliefs, drawing on the participants’ individual experience. Far from encouraging disinformation, AI, by adapting to individuals’ lived experiences, can therefore foster critical thinking and help people break free from dogma.
Besides, does the fact that this article was written with the help of artificial intelligence make it any less relevant? Does this method of production call into question the credibility of the ideas developed in it? If not, why should we be uneasy about a tool that, when used well, improves our ability to express ourselves? Is it really its use that is the problem, or rather the idea we have of it? Therefore, instead of seeing AI as a threat to critical thinking, should we not regard it as a lever for deeper thinking and intellectual emancipation? In that case, could it not be a lever for accelerating fair and informed transitions?
References
[1] Journet, N. (2005). ‘L’affaire Sokal : pourquoi la France ?’. Sciences Humaines. Reprinted in C. Halpern, F. Cusset and N. Journet (eds), Pensées rebelles. Foucault, Derrida, Deleuze (pp. 19–25). Éditions Sciences Humaines, 2013. doi.org/10.3917/sh.colle.2013.02.0019
[2] Silberzahn, P. (2 October 2023). ‘Et si l’IA était la chance des moins qualifiés ?’. Philippe Silberzahn [blog]. philippesilberzahn.com/2023/10/02/et-si-l-ia-etait-la-chance-des-moins-qualifies/
[3] Bègue-Shankland, L. (7 December 2020). ‘Phébé – Pourquoi nous prêtons plus attention au négatif qu’au positif’. Le Point. lepoint.fr/monde/phebe-pourquoi-nous-pretons-plus-attention-au-negatif-qu-au-positif-07-12-2020-2404532_24.php
[4] Silberzahn, P. (13 November 2023). ‘Les sept raisons pour lesquelles vous êtes (déjà) en train de rater le virage de l’IA’. Philippe Silberzahn [blog]. philippesilberzahn.com/2023/11/13/les-sept-raisons-pour-lesquelles-vous-etes-deja-en-train-de-rater-le-virage-de-l-ia/
[5] Ritchie, H. (18 November 2024). ‘What’s the impact of artificial intelligence on energy demand?’. Sustainability by Numbers. sustainabilitybynumbers.com/p/ai-energy-demand
[6] Goldman Sachs Research (2024). GS SUSTAIN: AI/data centers’ global power surge: The push for the ‘Green’ data center and investment implications. goldmansachs.com/insights/goldman-sachs-research/the-push-for-the-green-data-center
[7] Howarth, N., Camarasa, C., Lane, K. and Risquez Martin, A. (21 July 2023). ‘Keeping cool in a hotter world is using more energy, making efficiency more important than ever’. International Energy Agency (IEA). iea.org/commentaries/keeping-cool-in-a-hotter-world-is-using-more-energy-making-efficiency-more-important-than-ever
[8] Silberzahn, P. (25 November 2024). ‘Cinq principes pour réglementer (intelligemment) l’IA’. Philippe Silberzahn [blog]. philippesilberzahn.com/2024/11/25/cinq-principes-pour-reglementer-intelligemment-l-ia/
[9] Boudry, M. (11 April 2021). ‘Il est temps d’enterrer le principe de précaution’. Le Point. lepoint.fr/debats/il-est-temps-d-enterrer-le-principe-de-precaution-11-04-2021-2421671_2.php
[10] Piccolo, C. and Gouthière, F. (8 February 2025). ‘“J’ai écrit 20 lignes et ChatGPT m’a fait les 40 autres” : comment l’IA modifie les pratiques des journalistes’. Libération. liberation.fr/checknews/jai-ecrit-20-lignes-et-chatgpt-ma-fait-les-40-autres-comment-lia-modifie-les-pratiques-des-journalistes-20250208_KRQVNECQGRCVNLGZO2JUXCC45I/
[11] Buéno, A. (21 April 2023). ‘IA, la nouvelle machine à écrire, par Antoine Buéno’. Libération. liberation.fr/international/ia-la-nouvelle-machine-a-ecrire-par-antoine-bueno-20230421_HIIYGK3WJNEMTHMFDBUTV26QWA/
[12] Galibert, W. (2023). ‘Quand l’écrivain Hervé Le Tellier affronte ChatGPT : l’IA va-t-elle tuer la littérature ?’. RTL. rtl.fr/actu/international/quand-l-ecrivain-herve-le-tellier-affronte-chatgpt-l-ia-va-t-elle-tuer-la-litterature-7900483834
[13] Buéno, A. (2 April 2024). ‘Les deepfakes, une révolution spirituelle ?’. L’Opinion. lopinion.fr/tribune/antoine-bueno-les-deepfakes-une-revolution-spirituelle
[14] Samuel, S. (16 April 2025). ‘The real argument artists should be making against AI’. Vox. vox.com/artificial-intelligence/408786/ai-art-studio-ghibli-moral-injury-copyright
[15] Gadsby, S. and Van de Cruys, S. (12 September 2024). ‘The surprising role of deep thinking in conspiracy theories’. Psyche. psyche.co/ideas/the-surprising-role-of-deep-thinking-in-conspiracy-theories
[16] Costello, T. H., Pennycook, G. and Rand, D. G. (2024). ‘Durably reducing conspiracy beliefs through dialogues with AI’. Science, 385(6714), eadq1814. doi.org/10.1126/science.adq1814
0 comments