 Intriguing study by MIT Professor Markus Buehler, who presents original ways in which AI technologies can help accelerate scientific discovery through knowledge graphs, creative thinking, unterdisciplinary approaches, including artistic inspiration, and philosophical insights. More precisely, Buehler, 2024, constructed knowledge graphs mapping the knowledge contained in 1,000 scientific papers on biological materials and tested AI’s ability to:
🌹 Extract traversal paths, i.e., identify connections between seemingly unrelated ideas, such as the hydrophobic properties of rose petals and nacre-inspired cement. 🎵 Perform isomorphism analysis, i.e., explore structural similarities across distinct knowledge domains. A comparison between the patterns of relationships underlying biological materials and Beethoven’s 9th Symphony thus suggested the existence of universal design principles governing emerging properties in complex systems. 👩🎨 Leverage multimodal capabilities to find inspiration for scientific research in artistic works. Analyzing Kandinsky’s “Composition VII”, vision-equipped generative AI models were thus able to “draw inspiration from the abstract and emotional qualities of expressionist art to guide the development of sustainable mycelium-based materials.” 🤔 Draw philosophical parallels and conclusions from these experiments, including references to Gilles Deleuze’s “flat ontology” and “radical immanence”. Fascinated with Leibniz’ metaphysics since my late teens, I couldn’t help but smile when I read Buehler’s suggestion at the book of nature is written in computational language: “Extending these ideas, the isomorphisms spanning physical, biological, and artistic domains may be phenomenological reflections of reality’s deepest, unifying metaphysical principles - elemental dispositions and generative regimes rooted in symmetries and computational motifs whose abstract essences precede and govern all existent complexities as their ultimate ‘source code’. This would mean that at the most fundamental level, reality may be governed by basic patterns or rules that are computational in nature, akin to how a computer program operates based on its source code, and that these patterns are characterized by symmetry and generative processes that produce the complex world we observe”. To be clear, some aspects of the methodology and findings of this particular study give me pause, but the general idea that AI can be used to enhance creative thinking and help us connect distant ideas and domains is an important one. Original study: https://arxiv.org/pdf/2403.11996.pdf
0 Comments
Your comment will be posted after it is approved.
Leave a Reply. |