Institut Jean Nicod

Accueil > Séminaires/Colloques > Colloques et conférences > Colloques et Workshops > Workshop - Créativité, Imagination et Intelligence Artificielle



Workshop - Créativité, Imagination et Intelligence Artificielle

Date et lieux : Mercredi 19 (salle Borel) & Jeudi 20 (Salle Berthier) Mars, ENS, 29 rue d’Ulm, 75005 Paris

Organisateurs  : Margherita Arcangeli, Séphora Bennaoum & Jérôme Dokic

 
Cette conférence, organisée dans le cadre du projet PSL Global Seed Fund CreAItion (« CREativité et imaginaTION par le prisme de l’Intelligence Artificielle »), vise à aborder sous différents angles (philosophie de l’esprit, philosophie du langage, philosophie de l’intelligence artificielle, théorie et pratique de l’art, sciences cognitives) les questions relatives à la créativité, à l’imagination et à leurs liens à la lumière de la prolifération de systèmes artificiels capables de créer (par exemple, des poèmes ou des peintures). Voici quelques-unes des questions les plus pressantes : Peut-on réellement parler de créativité dans ces contextes ? S’agit-il de créativité sans imagination ? Des processus non créatifs peuvent-ils donner naissance à des produits créatifs ? Ces systèmes n’ont-ils pas une certaine forme d’imagination ? Dans quelle mesure les modèles artificiels peuvent-ils être utilisés pour supporter la créativité humaine plutôt que de la supplanter ?

Intervenants confirmés :

  • Daniel Andler (Sorbonne Université - ENS/DEC)
  • Gregory Bochner (Collège de France)
  • Simona Chiodo (PoliMI)
  • Dimitri Coelho Mollo (Umeå University)
  • François Garnier (ENSAD) 
  • Steve Humbert-Droz (Umeå University)
  • Julia Langkau (University of Geneva)
  • Piera Maurizio (Institut Jean-Nicod)
  • Andrea Rivadulla Durò (Universitat Autónoma de Barcelona)
  • Emmanuelle Volle (Paris Brain Institute)
PROGRAMME :

Mercredi 19 mars

09:30 – 10:30 Simona Chiodo (PoliMI), “What creativity is always for (from human to artificial ?)”
short break
10:45 – 11:45 Steve Humbert-Droz (Umeå University), “Is Creative Thinking Imagination or Intelligence ?”
11:45 – 12:45 Emmanuelle Volle (Paris Brain Institute), “Neurocognitive processes of creative thinking : from memory to decision-making”

long break

14:30 – 15:30 Andrea Rivadulla Durò (Universitat Autónoma de Barcelona), “Creative Transitions”
15:30 – 16:30 Julia Langkau (University of Geneva), “Creative imagining as guided mind-wandering”
short break
16:45 – 17:45 Piera Maurizio (IJN), “Open-mindedness and Artificial Intelligence”
17:45 – 18:45 François Garnier (ENSAD), “Stimulate creativity, not simulate it”


Jeudi 20 mars

09:30 – 10:30 Daniel Andler (Sorbonne Université - ENS/DEC), “Creativity, problem-solving and artificial intelligence”
short break
10:45 – 11:45 Gregory Bochner (Collège de France), “From strong AI to the ‘hard’ problem of other minds”
11:45 – 12:45 Dimitri Coelho Mollo (Umeå University), “What could AI ’imagination’ possibly look like ?”

 

RESUMES

Daniel Andler (Sorbonne Université - ENS/DEC), “Creativity, problem-solving and artificial intelligence”

Creativity is often closely connected to problem-solving. From art history and science to management and everyday life, it is argued (1) that creativity is required to solve difficult problems and (2) reciprocally, creative achievements, in art where it isn’t obvious as well as science where it is, turn out to be solutions to some problem. Following this line, it is tempting to conclude that artificial intelligence, with its seemingly limitless problem-solving capacities, can be creative. I will show that this line of thought is deeply mistaken, resting as it does on an equivocation regarding the notion of a problem. Artificial intelligence is at most capable of an Ersatz of creativity.

 

Gregory Bochner (Collège de France), “From strong AI to the ‘hard’ problem of other minds”

In the work of David Chalmers, we find the following claims : there is a ‘hard’ problem of consciousness which undermines materialism ; we need not be worried by the problem of other minds in our world ; and the strong AI thesis is true. I will argue that this combination is not legitimate : if there is a hard problem of consciousness, then we have good reasons to worry about other minds in our world, and the right attitude about the prospects of strong AI becomes skepticism or agnosticism. My argument will rely on the conceivability of partial zombie worlds : worlds in which some individuals are conscious while others are zombies.

 

Simona Chiodo (PoliMI), “What creativity is always for (from human to artificial ?)”

Can an artificial intelligence (AI) that is greatly autonomous from humans’ inputs be thought of as creative ? I will answer in the negative, starting with the analysis of the case of Edmond de Belamy, together with further cases. More precisely, I will offer the following three kinds of arguments. First, what results from testing the idea of creative AI against the background of traditional definitions of creativity, especially from a philosophical perspective. Second, what results from testing the idea of creative AI against the background of the notion of meta-sensemaking. Third, what results from testing the idea of creative AI against the background of the epistemology of creativity. After arguing that the possible answer to my starting question is negative, at least in the sense that speaking of creative AI means speaking of something far different from human creativity, I will argue that the reflection upon the idea of creative AI should not make us lose sight of what is most important : what creativity is always for, i.e. making sense of our own existence, from art to anything, not only because it is a kind of existential urgency for us but also because it is pleasant in itself for us.

 

Dimitri Coelho Mollo (Umeå University), “What could AI ’imagination’ possibly look like ?”

Imagination seems to be an important if not crucial capacity underlying the richness, flexibility and generality of human cognition and intelligence. It is plausible that, if AI systems are ever to become truly intelligent, they will need to be endowed with an analogous capacity. However, the nature and roles of human imagination are themselves contested ground.
In this talk, in light of recent work on human imagination, we will suggest preliminary, basic criteria for what AI systems would have to possess in order to approximate the functional significance of imagination in human cognition. We will examine a few case studies, indicating what they get right and what they still miss in implementing quasi-imaginative capacities in AI systems.

 

François Garnier (ENSAD), “Stimulate creativity, not simulate it” 

À travers la présentation de deux projets de recherche création, associant réalité virtuelle et intelligence artificielle, réalisés dans le cadre du groupe de recherche Spatial Media de l’EnsadLab (Tamed Cloud, interaction sensible avec de grand corpus d’images ; The immersive rehearsal project, réalisé par Alex Belin de l’Ensad et Salomé Gobbi de Mines Paris avec une thèse binômée PSL), nous nous intéresserons au rôle que peut prendre l’intelligence artificielle dans un processus de création artistique.
S’il nous paraît fondamentalement stérile de tenter de simuler l’acte artistique, l’œuvre déshumanisée perdant son statut d’œuvre, comment l’intelligence artificielle peut-elle être utilisée en tant qu’outil stimulant l’acte créatif, en mettant en condition ou en assistant l’artiste.

 

Steve Humbert-Droz (Umeå University), “Is Creative Thinking Imagination or Intelligence ?”

In philosophy, creativity — and, more specifically, creative thinking (as opposed to creative achievements) — is often associated with imagination (e.g., Currie & Ravenscroft 2002 ; Hills & Bird 2019 ; Picciuto & Carruthers 2014 ; Stokes 2014). Even scholars who reject a substantial connection between creativity and imagination take this relationship as a starting point for their inquiry. In psychology, however, creative thinking is frequently associated with intelligence (e.g., Guilford 1967 ; Kaufman 2009 ; Sternberg 1985 ; Stevenson et al. 2021).
In this talk, I explore whether creative thinking substantively belongs more to the category of imagination or intelligence. Is creative thinking a sui generis type of imagination or intelligence ? Can it be reduced to a known type of imagination or intelligence ? Or is imagination or intelligence a necessary (but not sufficient) component of creative thinking ?
I argue that creativity does not neatly fall within either category, but while it fits into the imagination category only when imagination is taken in a trivial sense, it fits into the intelligence category in a slightly more fruitful way. I conclude that philosophers should analyze creativity with more emphasis on its relationship with intelligence than with imagination.

 

Julia Langkau (University of Geneva), “Creative imagining as guided mind-wandering”

In this paper, I will defend an account of creative imagining as a form of mind-wandering. I will argue that creative imagining is driven by the wandering mind and guided by attention to experiential value. It is outward-directed, sometimes involves an aspirational goal and, as such, differs from daydreaming and fantasy. With this notion of creative imagining at hand, we can distinguish cases of creative thinking with intrinsic value from cases of creative thinking without such value.

 

Piera Maurizio (Institut Jean-Nicod), “Open-mindedness and Artificial Intelligence”

Open-mindedness requires, among other things, being attuned and responsive to relevant information when inquiring into a subject matter. In this talk, I specify this requirement by examining the norms that govern how we direct attention while gathering evidence. I propose that effective inquiry depends on meeting at least two key conditions :

1. A well-functioning priority mechanism – Psychologically and epistemically, one’s attention should be attuned to the subject matter of inquiry. At a minimum, this means that one’s search mechanisms align with relevant features in the environment, based on how the individual sets and understands their inquiry goals. This attunement ensures a disposition to notice pertinent information.

2. Sensitivity to new and unexpected evidence – Beyond initial attunement, a well-functioning epistemic agent must also be dynamically responsive. This means being disposed, at the individual level, to recognize and appropriately react to new or unforeseen evidence that bears on the subject of inquiry.

In the final part of the talk, I explore whether Artificial Intelligence can exhibit open-mindedness in this sense.

 

Andrea Rivadulla Durò (Universitat Autónoma de Barcelona), “Creative Transitions”

Most of the time, thought follows well-worn routes : logical inferences unfold, memories are triggered, and perceptual simulations take shape. But occasionally an unusual leap occurs— less constrained by rules and past experience, more like opening a new space. Intuitively, these leaps form a core part of creative thought. But what distinguishes a creative transition from other kinds of transitions in thought ? More specifically, this paper addresses the following question : given a specific input or initial state, what features of the transition leading to an output render it creative ?
Process-based accounts of creativity agree that creative thought is inherently non-formulaic—it is not governed by rigid, rule-bound procedures (Gaut, 2003 ; Brainard, 2024). Consequently, a logical inference, which strictly follows from its premises to its conclusion, cannot be creative because it leaves no room for spontaneity. Yet in explaining other aspects of creative transitions—such as the cognitive sources of novelty—process-based accounts emphasize more controversial features such as stochasticity (Carruthers, 2020) or associative processes (Stokes, 2007)—thereby downplaying the roles of rationality and agency in creative transitions. In response, drawing on accounts of inferential and associative transitions (Quilty-Dunn & Mandelbaum, 2018 ; Quilty-Dunn & Mandelbaum, 2019 ; Shea, 2024), I propose a novel framework that integrates rationality and agency into our understanding of creative transitions.
My aim in this paper is twofold. First, I critically assess two perspectives that minimize the roles of rationality and agency in creative transitions : (1) those that attribute creativity primarily to associative processes (Mednick, 1962 ; Stokes, 2007 ; Beaty & Kennett, 2023) and (2) those that emphasize stochastic mechanisms (Simonton, 2003, 2004 ; Carruthers, 2020). Second, I develop an empirically informed account of the cognitive operations underlying creative transitions and delimit the conditions under which these transitions can rightfully be considered creative.

 

Emmanuelle Volle (Paris Brain Institute), “Neurocognitive processes of creative thinking : from memory to decision-making”

Creative thinking orchestrates the production of ideas and solutions that are both original and effective, enabling us to adapt to change, solve problems, and drive innovation. Psychological and neuroimaging research supports a dual-process model in which creativity relies on two key components : idea generation and evaluation, each involving distinct yet interconnected brain networks. The generation component produces candidate ideas, while the evaluation component assesses their originality and adequacy. However, the precise neurocognitive mechanisms underlying these processes—and how they lead to creative ideas—remain to be fully understood. I will present recent findings that illuminate some of these mechanisms. Specifically, I will explore how the structure of semantic memory and memory search processes influence idea generation, the importance of valuation in idea evaluation, and their interaction with cognitive control.

 

Pour plus d’information, veuillez contacter Margherita Arcangeli


CNRS EHESS ENS ENS