Editorial illustration for Birkhoff’s 1930s ‘measure’ and AICAN’s ‘novelty’ probe AI aesthetics
Birkhoff’s 1930s ‘measure’ and AICAN’s ‘novelty’ probe...
The MIT Keller Gallery will host “Beyond Data‑Driven Aesthetics” through June 30, a show that pulls together philosophy, mathematics, computer science and design computation into tangible installations and interactive visualizations. Curated by Alexandros Haridis, an MIT Architecture alumnus and researcher, the exhibition asks how computing can become a medium for creative production and aesthetic judgment in architecture and the applied arts.
While finishing his PhD in design and computation at MIT’s Department of Architecture around 2022, Haridis watched systems such as ChatGPT and Stable Diffusion surge into public debate over creativity, design and even high‑profile art auctions. That moment sparked the first of three research strands feeding the show.
A second strand looks back to the 1956 Dartmouth Summer Research Project, where creation and evaluation were singled out as one of seven key dimensions of human intelligence that AI should tackle.
A third line, hinted at but not fully detailed here, stems from design‑computation work on shape grammars that probes the interplay between human insight and algorithmic processes. Together, these threads frame a historical and technical inquiry into what it means to judge beauty when machines are part of the equation.
For example, "measure" refers to mathematician George Birkhoff's work in the 1930s to quantify aesthetic value mathematically, while "novelty" examines how the machine learning system AICAN judges generated images according to a theory in cognitive aesthetics that balances familiarity and deviation from known artistic styles.
Across all five cases, the key insight is that design itself can function as a method of interpretative translation -- a way of making visible, tangible, and experiential what traditional academic scholarship in technical domains typically communicates only through words and word-like representational devices, such as scientific diagrams and tables.
Q: What questions are you hoping to explore next?
A: "Beyond Data-Driven Aesthetics" is conceived both as a research exhibition and as an ongoing platform for investigating how computational systems participate in processes of aesthetic judgment, generation, and transformation across architecture and the applied arts.
One of the central questions of the exhibition -- and one that researchers across architecture, design, and engineering are increasingly focusing on -- is computational evaluation beyond purely performative or functional requirements.
Why this matters Can a machine truly gauge beauty? A question, not an answer. The MIT Keller Gallery’s “Beyond Data‑Driven Aesthetics” pulls together century‑old math and today’s generative models, reminding us that the quest to codify taste is far from settled.
We see Birkhoff’s 1930s “measure” resurfacing as a historical touchstone, while AICAN’s “novelty” metric attempts to balance familiarity with deviation in generated images. For developers, the exhibit offers a concrete illustration of how aesthetic theory can be embedded in code, yet the translation from gallery to product remains ambiguous. Founders may wonder whether such quantifications can justify commercial design tools; researchers can note that the installation makes explicit the assumptions behind current loss functions.
Still, it is unclear whether a single numerical score can capture the nuance of architectural judgment or user preference across cultures. As we watch these ideas migrate from museum walls to development pipelines, we should keep a healthy doubt about their scalability and relevance beyond controlled settings.
Further Reading
- Papers with Code Benchmarks - Papers with Code
- Chatbot Arena Leaderboard - LMSYS
Related Reading
Research & Benchmarks Google's FACTS benchmark shows 70% factuality ceiling across four tests
Research & Benchmarks Databricks finds multi-step agents beat single-turn RAG by 21% to 38% on STaRK
Research & Benchmarks Nvidia's DLSS 4.5 beta adds 6x Multi Frame Generation for RTX 50 GPUs
From Archives MiniMax Token Plan offers extensive coding model access for USD 20/month
From Archives Sina's VibeThinker-3B probes limits, shows reasoning compresses, knowledge weak
Common Questions Answered
What is Birkhoff's 1930s 'measure' and how does it relate to the MIT exhibition?
Birkhoff's 1930s 'measure' is a mathematical approach to quantifying aesthetic qualities that serves as a historical touchstone in the 'Beyond Data-Driven Aesthetics' exhibition at MIT Keller Gallery. The exhibition resurrects this century-old mathematical concept to demonstrate how computing can be applied to aesthetic judgment and creative production in architecture and design.
How does AICAN's 'novelty' metric function in evaluating generated images?
AICAN's 'novelty' metric attempts to balance familiarity with deviation in generated images, creating a framework for assessing creative output by generative models. This approach represents a contemporary effort to codify aesthetic taste by measuring how well an AI-generated image maintains recognizable elements while introducing novel variations.
What is the main curatorial question posed by 'Beyond Data-Driven Aesthetics'?
The exhibition, curated by Alexandros Haridis, asks how computing can become a medium for creative production and aesthetic judgment in architecture and the applied arts. It explores the fundamental question of whether machines can truly gauge beauty by integrating philosophy, mathematics, computer science, and design computation into tangible installations and interactive visualizations.
Why does the exhibition suggest that codifying taste remains unsettled?
The exhibition demonstrates that the quest to quantify aesthetic judgment through mathematical measures and AI metrics is far from complete and conclusive. By juxtaposing Birkhoff's historical mathematical approach with modern generative models like AICAN, it illustrates that defining and measuring beauty remains an open-ended philosophical and computational challenge.
Further Reading
- Papers with Code Benchmarks — Papers with Code
- Chatbot Arena Leaderboard — LMSYS
