[CESERH] CI La généalogie des formes dans la perspective du Deep Learning 8/11/2019 [1/9]

[CESERH] CI La généalogie des formes dans la perspective du Deep Learning 8/11/2019 [1/9]

[CESERH] CI The genealogy of forms from the Deep Learning perspective 8/11/2019 [1/9]

by Pierluigi Basso Fossali · View the channel on PSI

Humanities, Social Sciences & Thought Computing & Cybersecurity UXApplied computingUXADigital Humanities
🎙 Pierluigi Basso Fossali 👥 178 📅 November 19, 2019 ⏱ 51 min 👁 111 📄 expert opinion 🧭 2026-10-10
Available in: English (current) Français

Keywords

semioticsvisual languageformgenealogydeep learning

Summary

Pierluigi Basso Fossali opens with a tribute to Simondon’s early call to integrate informatics into cultural sciences, arguing that digitalization can unlock latent potentials of meaning in images. He then frames the talk as an exploration of how visual language, unlike verbal language, resists stable institution and relies on local negotiation. He proposes a comparative genealogy of forms, drawing on examples from painting, cinema, and music, and outlines categories such as transformation families, transposition between media, and issues of complementarity. The discussion shifts to the idea of form as a resilient identity that resists decomposition, and to the necessary tension between tradition and innovation in any semiotic practice. He addresses clichés and stereotypes, distinguishing their cultural weight from their artistic uses, and emphasizes the need for a critical and ethical approach to visual pollution. Throughout, he references Focillon, Deleuze, and others to ground his arguments in art theory. The talk concludes by advocating for a dynamic, generative view of forms, where genealogies are always counterbalanced by counter-movements, and where deep learning serves as a vague horizon rather than a concrete methodology.

181 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of this talk lies in its rich conceptual web, linking semiotics, art history, and digital perspectives. The argumentation is coherent but highly speculative, relying on analogies and philosophical references rather than empirical evidence. The speaker’s expertise is evident in the breadth of examples, though the deep learning connection is never concretely developed, which weakens the overall impact.

Scientific Rigor, Source Quality, Title Accuracy

The talk demonstrates strong scholarly rigor in its engagement with semiotic traditions (Simondon, Focillon, Deleuze), but no formal sources or citations are provided in the description, making verification difficult. The title’s promise of a deep learning perspective is only loosely addressed, as the focus remains on traditional humanities methodologies. The content is therefore intellectually stimulating but not rigorously anchored to current AI research.

137 words

Title / Content Match

The title suggests a deep learning perspective, but the talk focuses on semiotics and art theory; the deep learning reference is peripheral. Nonetheless, the title does mention genealogy of forms, which is the core topic.

Quality & Reliability

6/10

The talk is a scholarly exposition of semiotic theory with references to thinkers like Simondon, Focillon, and Deleuze, but it lacks supporting empirical evidence or explicit citations to specific studies, and the connection to deep learning remains metaphorical.

Key Moments

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

Cited Sources

Contribution & Novelties

The talk offers a novel synthesis of semiotic theory and digital culture, proposing that deep learning models might be viewed as heuristic devices for mapping visual genealogies. It emphasizes the dynamic, unstable nature of forms and calls for a comparative approach to visual meaning-making.

Pour aller plus loin :

98 words

Radar Profile

The scores reveal a talk that is dense in content (quantity 7) but only moderately rigorous in method (quality 6), with a low technical level (5) and moderate reliability (6). The radar suggests a strong conceptual breadth but limited empirical grounding, typical of a humanities lecture engaging with technological themes only superficially.

Reliability 6/10