Peut-on produire du sens à l'époque des GAFAM ?

Value of the Information & Strength of the Argument

The value of the information lies in its conceptual framework rather than in new empirical findings. Vitali-Rosati offers a clear and provocative thesis: thought is inseparable from its material support, so the choice of a file format, a server, or a videoconferencing protocol has the same kind of impact on what can be thought as the choice to delegate writing to ChatGPT. This reframing is valuable because it dissolves the artificial hierarchy between ‘important’ AI issues and ‘minor’ technical details. The argumentation is structured around a few strong oppositions: neutral tools vs. situated values; solutionism vs. critical questioning; false literacy (adapting to tools) vs. true literacy (questioning one’s needs and values); high-tech vs. low-tech bricolage. The examples are concrete and effective: ChatGPT as a search engine, WhatsApp turning photos into an endless flux, Word/ .docx shaping humanities writing, Zoom vs. Jitsi vs. BigBlueButton embodying different values of encounter, and the sailor who can navigate by alignments rather than GPS. The reasoning is coherent and rhetorically well built, but it remains largely assertive: the causal claim that tools determine thought is illustrated rather than demonstrated, and counterarguments (e.g., that users can repurpose tools, or that some standardization enables collective work) are not seriously engaged. The talk is thus a strong opinion piece with heuristic value, not a proof.

Scientific Rigor, Source Quality, Title Accuracy

Scientific rigor: the talk is an expert opinion delivered in an academic setting (Sens Public, Paris, January 23, 2025). It draws on a long intellectual lineage—Leibniz, Pascal, Raymond Lull, Turing’s 1950 ‘Computing Machinery and Intelligence’, Marx on capitalism, Barlow’s 1996 Declaration of the Independence of Cyberspace—but these are invoked from memory, without precise references or quotations. No data, no methodology, and no systematic review are presented. The claims about the convergence of GAFAM and state power are supported by a single visual example (the Trump administration) and general observation, not by documented analysis. Quality of sources: the transcript contains no explicit citations with titles, authors, or URLs; the only named works are the speaker’s own book ‘Les logiques du bug’ (La Découverte, 2024) and Turing’s article. The description provides no links. The talk is therefore rich in conceptual references but poor in verifiable sourcing. Adequation title/content: the title ‘Peut-on produire du sens à l’époque des GAFAM ?’ accurately reflects the central question addressed throughout the talk.

The title directly matches the content: the talk addresses whether meaning can still be produced under the dominance of GAFAM platforms and their convergence with state power.

7/10

The talk is a coherent expert synthesis by a recognized researcher in digital humanities, grounded in a long intellectual tradition (Leibniz, Pascal, Turing, Marx, cyberlibertarianism). It offers conceptual clarity and concrete examples, but remains an argumentative essay rather than an empirical study; no quantitative data, no systematic literature review, and no explicit citations with references are provided in the transcript.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 01:00Introduction: the end of the idealistic phase of the internet and the capture of digital spaces by corporations.
• 03:00The convergence between GAFAM and state power, illustrated by the Trump administration and tech leaders.
• 05:00Central thesis: thought is always a material inscription; the format, server, and protocol shape what can be thought.
• 10:00Critique of the 'solutionist' rhetoric: neutral tools, single scale of performance, naturalized values.
• 15:00The imperative of productivity and the functional imperative as the core of the dominant worldview.
• 20:00Examples of aberrant uses: ChatGPT as a search engine, Word/.docx for humanities, WhatsApp as an archive.
• 25:00False literacy (adapting to tools) vs. true literacy (questioning one's needs and values).
• 30:00The principle of bricolage and low-tech: high competence correlates with simpler technologies; the sailor and GPS example.
• 35:00Critique of the term 'artificial intelligence' as a vague media keyword, and of 'generative' as conceptually empty.
• 40:00Historical perspective: AI debates from Leibniz, Pascal, and Lull to Turing's 1950 paper.

• Les logiques du bug — Book by Marcello Vitali-Rosati (La Découverte, 2024), mentioned as the source of his analysis of the convergence between states and tech corporations.
• Computing Machinery and Intelligence (Turing, 1950) — Referenced as the founding article of artificial intelligence, dated 1951 in the talk.
• A Declaration of the Independence of Cyberspace (John Perry Barlow, 1996) — Cited as the emblematic expression of the idealistic vision of the early internet.

• Evgeny Morozov, To Save Everything, Click Here (2013) — Critique of solutionism and technological neutrality, echoing the talk's central argument.
• Shoshana Zuboff, The Age of Surveillance Capitalism (2019) — Analysis of the economic and political power of GAFAM, consistent with the talk's diagnosis.
• Lawrence Lessig, Code and Other Laws of Cyberspace (1999) — Thesis that code regulates behavior and embodies values, closely related to the materiality argument.

• Technological determinism — The talk explicitly rejects technodeterminism while simultaneously claiming that tools determine what can be thought, which some critics would see as a tension or an implicit determinism.
• Instrumentalism / neutral tools — The view that technologies are neutral instruments serving user ends is directly contradicted by the talk; defenders of this position would argue that users can repurpose tools and that the speaker underestimates agency.