MIT Exhibition Exposes the Century-Old Quest for Computational Aesthetics

Can algorithms judge beauty? MIT's latest exhibition argues this isn't just an AI-era question, it's a century-old pursuit with surprising depth and ongoing relevance.

'Beyond Data-Driven Aesthetics,' on view at MIT's Keller Gallery through June 30, cuts through the noise surrounding AI-generated art. Curator Alexandros Haridis, an MIT Architecture alumnus, transforms dense algorithms and abstract theories into tangible installations, revealing how computing has long grappled with creativity and aesthetic judgment.

The Long Shadow of Computational Creativity

While ChatGPT and Stable Diffusion dominate headlines, Haridis argues that today's 'breakthroughs' in machine creativity often rehash decades-old questions. He points to the 1956 Dartmouth Summer Research Project, a foundational AI event, which already identified creation and evaluation as key targets for artificial intelligence.

This historical context is crucial. It suggests that current AI art tools, despite their outputs, may be treading well-worn philosophical ground rather than truly novel territory.

Algorithms Made Tangible

The exhibition's core strength lies in its translation of abstract computational concepts into physical, interactive experiences. Haridis organizes the show around five themes:

1. Aesthetic Measure
2. Aesthetic Guidelines
3. Algorithmic Aesthetics
4. Aesthetic Appropriation
5. Aesthetic Novelty

Each theme unpacks a specific approach to computational aesthetics. 'Aesthetic Measure,' for instance, explores 1930s mathematician George Birkhoff's attempts to quantify beauty mathematically. 'Aesthetic Novelty' examines how the AICAN machine learning system judges generated images, balancing familiarity against stylistic deviation.

By making these ideas tangible, the exhibition offers rare insight into the inner workings of aesthetic AI, moving beyond typical 'black box' presentations.

Design as Research Method

Haridis's approach reveals an intriguing methodological insight: design itself can serve as a powerful interpretative tool. By reconstructing algorithms, visualizing data, and creating physical artifacts, the exhibition makes visible what's often buried in dense academic papers.

This translation from text to experience isn't just about accessibility. It suggests new ways of conducting and communicating research, especially in technical fields traditionally reliant on words and diagrams alone.

From Art to Architecture: Computational Evaluation's Broader Impact

While focused on aesthetics, the exhibition's implications stretch further. Haridis points to growing interest in computational evaluation beyond purely functional requirements, applicable to buildings, structural forms, and everyday objects.

This raises a critical question: How can computation, whether rule-based or data-driven, inform our understanding of what contributes positively to human experiences in built environments?

A Necessary Historical Lens

'Beyond Data-Driven Aesthetics' offers a vital counterpoint to breathless AI art headlines. By excavating the long history of computational approaches to creativity, it provides much-needed context for today's developments.

The exhibition's value lies not in predicting AI's creative future, but in revealing the depth of thought already applied to these questions. For anyone grappling with AI's role in art and design, this historical perspective is invaluable, a reminder that today's 'innovations' often recycle decades-old inquiries in new technological packaging.

As AI tools become increasingly accessible to creators, understanding this lineage becomes crucial. It allows us to approach new developments with informed skepticism, recognizing both genuine advances and repackaged ideas. In doing so, we can better navigate the complex intersection of computation, creativity, and human experience.