Impossible to avoid it: over the past two years, conferences and debates on AI and its impact on creative industries have multiplied — media, music, design, luxury, advertising...
Faced with this flood of contradictory opinions, I am offering you an overview of what science says on the subject — the sciences: cognitive psychology, neuroscience, sociology of technology. A way to overcome the emotion of the debate and anchor the discussion in concrete facts.
Here is episode 2 out of 4 where we talk about the “average” effect.
Did you miss episode 1? Discover on this previous article the scientific components of creativity, and what that tells us about the AI vs Human match.
The risk of homogeneity: the last line of defence?
Researchers in the sociology of technology, such as Sherry Turkle (professor and renowned researcher at MIT in Boston), warn against the social implications of using AI in creation: the industrialization of creation via AI could lead to the standardization of culture and tastes.
This is one of the main concerns raised by researchers in the sociology and ethics of AI currently: the risk of standardization and depersonalization of creations. What specialists call the “average” effect.
The question of authenticity
Let's take a concrete example. When it comes to consumer products, studies in social psychology show that consumers attribute significant emotional value to products when they perceive authenticity behind them, something particularly true in industries such as luxury goods.
This “authenticity” is defined as intrinsically linked to a story, to the vision of a creator and to the soul of a work, especially when it comes to something new, new, surprising. Of creative.
In the luxury sector, for example, an industry for which I speak very often as a creativity speaker and an innovation speaker, the question arises: in a sector where uniqueness is the key to perceived value, if everything becomes predictable and algorithmic, where will be the place for Nonconformism and innovation, for personal creativity?
If AI creates objects based solely on algorithms, it risks producing works or objects that lack emotional depth.
At least that's what some people want to believe.
But the subject is not unanimous...
Research on how to overcome “averageism”
Current work aimed at going beyond the simple generative “averaging” of AIs is progressing, and progressing rapidly. Research in AI labs aims to go beyond the simple statistical imitation of AIs.
This is the case with those called the CAN - the Creative Adversarial Networks, which drive innovation by moving away from established styles, while models like those of Magenta From Google Brain produce music and art by integrating a context to explore new variants.
On the linguistic and textual level, OpenAI's GPT-4 already takes into account a “purpose” and a style, beyond pure statistical alignment.
At the same time, researchers such as Margaret A. Boden (Emeritus Professor of Cognitive Science at the University of Sussex, and founder of the School of Cognitive and Computing Sciences) emphasize in their research the need to “look for the unexpected” through exploration and self-evaluation mechanisms.
Finally, the path of reinforcement learning makes it possible to generate unanticipated behaviors, paving the way for more expressive and meaningful AIs.
Is emotion in AI generations coming soon?
The “average” effect is still one of the major limitations of AIs in generating a truly creative result, but current research shows that these systems are rapidly evolving towards greater sensitivity.
The ability to integrate a clearly expressed intention, emotion, or “purpose”, and even to move away from established patterns to think outside the box is no longer far away. A matter of time, really?
Episode 3: The legal limit: a real constraint for AI? , coming soon
***
This article is the 2nd in a series written in early 2025 devoted to the AI x Human match on the subject of creativity.
If you missed episode 1, watch it here: Episode 1: The Scientific Components of Creativity