# SoTA Canonical URL: https://jeanyoon.ch/oi/artworks/sota Type: artwork (multi-device web artwork) Artists: Jeanyoon Choi, Euan Kang, SeJoon Park, Minhyeok Seo, Prof. Yiyun Kang Year: 2024 Medium: Multi-Device Web Artwork (4 Displays, 2 Projectors, 1 Mobile, 4 Channel Audio) Korean page: https://jeanyoon.ch/oi/artworks/sota/ko ## Summary SoTA (State-of-the-Art) is a 2024 explainable-AI (XAI) multi-device web artwork created by Jeanyoon Choi with Euan Kang, SeJoon Park, Minhyeok Seo and Prof. Yiyun Kang at KAIST XD Lab. Using two projectors, four PCs and audience phones, it presents 118 neural network architectures in an interactive 3D, representational form (the Frontend). When audiences stop interacting, the screens switch to chaotic monochrome visuals of LLM token cross-similarities (the Backend). Drawing on Rich Sutton's 'Bitter Lesson' and modernism's turn to abstraction, it asks whether AI is drifting toward self-referential abstraction. It opened at KAIST Art Museum (December 2024 to June 2025). ## 요약 (Korean) 『SoTA』(2024)는 최정윤과 KAIST XD Lab 공동 제작자들이 만든 설명 가능한 AI(XAI) 멀티 디바이스 웹 아트워크로, 118개 신경망 구조를 3차원으로 탐색하게 한 뒤 LLM 토큰 유사성의 추상적 시각화로 전환되며 KAIST 미술관에서 전시되었습니다. ## Concepts defined - **Frontend and Backend (SoTA)**: The work's two phases: the Frontend presents neural architectures in a representational, interactive 3D form, while the Backend, triggered when audiences leave, shows abstract monochrome visuals of LLM token cross-similarities. ## Questions this answers ### What is SoTA by Jeanyoon Choi? SoTA (2024) is an interactive explainable-AI multi-device web artwork by Jeanyoon Choi and KAIST XD Lab collaborators. It lets audiences explore 118 neural network architectures in 3D and then shifts to abstract visualisations of LLM token similarities. ### How do audiences interact with SoTA? Visitors scan a QR code and use their phones to select and navigate neural network models shown across projectors and four screens. When they leave, the installation switches from this representational Frontend to an abstract monochrome Backend. ### Where has SoTA been exhibited? SoTA was shown in the KAIST Art Museum Opening Media Art Exhibition from 17 December 2024 to 30 June 2025, and at ACM DIS 2026 Interactivity at the National University of Singapore in June 2026. ### Is there a research paper about SoTA? Yes. 'SoTA: An Interactive Art Exhibition for Public AI Engagement' by Jeanyoon Choi, Intae Hwang, SeJoon Park, Hyungjun Cho, Heejae Bae and Yiyun Kang appeared at ACM DIS 2026 and received an Honourable Mention. ### Why was SoTA made as an AI art exhibition rather than an AI explainer? The accompanying DIS 2026 paper argues that public AI engagement needs more than technical explanation. SoTA uses immersive visualisation, embodied phone-based interaction and an artistic narrative ending in an unexpected crash to provoke affective and critical reflection; in a study with 33 visitors it produced curiosity, productive ambivalence and societal questions about AI. Topics: explainable ai, neural network visualisation, multi-device web artwork, abstraction, large language models, public ai engagement, ai art exhibition, public ai literacy References: [Explainable artificial intelligence](https://en.wikipedia.org/wiki/Explainable_artificial_intelligence), [Richard S. Sutton](https://en.wikipedia.org/wiki/Richard_S._Sutton), [Jackson Pollock](https://en.wikipedia.org/wiki/Jackson_Pollock), [Modernism](https://en.wikipedia.org/wiki/Modernism), KAIST Art Museum, Yiyun Kang, Euan Kang, SeJoon Park, Minhyeok Seo ## Description 『SoTA』 is an interactive XAI Multi-Device Web Artwork. Audience’s phone initially navigates a 3d visualisation of 118 neural networks; when they leave, all screens erupt into chaotic monochrome patterns, exposing AI’s drift from human comprehension toward “AI for AI’s sake”. State-of-the-Art (SoTA) refers to cutting-edge AI models, analogous to achieving an artistic pinnacle. Early SoTA architectures (CNN/RNN) were designed to mimic human cognition, echoing the representational approach. Yet, as Rich Sutton’s “Bitter Lesson” argues, ever-growing GPU power now eclipses human-designed intricacies, nudging SoTA from carefully crafted forms toward vast, computation-driven abstractions. 『SoTA』 is an XAI(Explainable AI)-driven interactive multi-device web artwork that makes our evolving metaphor tangible. Two projectors, four PCs, and audiences’ mobiles present 118 neural architectures in a 3D, interactive, and representational form—initially mirroring SoTA’s earlier pre-modernist era (the Frontend). But once audiences stop interacting and leave the site, unprecedentedly, the screens flip to chaotic monochrome visuals that abstractly depict LLM token cross-similarities (the Backend). Like Pollock’s expressive drips, the swirling lines of connectionism immerse audiences in a vortex of high-dimensional complexity no longer intelligible to us. Mirroring Modernism’s shift, 『SoTA』 reveals how AI can reshape society, identity, and creativity. It questions whether massive models will foster inclusive, human-centred progress or slip into self-referential abstraction—a Black Box. By uncovering AI’s hidden layers, 『SoTA』 urges us to reflect on the implications of State-of-the-Artificial Intelligence and responsibly guide these systems. ## Exhibitions - KAIST Art Museum Opening Media Art Exhibition, KAIST Art Museum, 17 December 2024 – 30 June 2025 (https://sota-xdlab.net/) - ACM DIS 2026 Interactivity, National University of Singapore, Singapore, June 2026 (https://dis.acm.org/2026/program-2026/) Artwork website: https://sota-xdlab.net/ Video: https://player.vimeo.com/video/1061512520?h=862a6a902a ## References - [XD Lab](https://www.xdlab.net/projects/sota) - [Interactive artwork](https://sota-xdlab.net/) ## Images - https://jeanyoon.ch/artworks/sota/1.webp — Installation view of 『SoTA』 in KAIST Art Museum. - https://jeanyoon.ch/artworks/sota/2.webp — The four screens and green projection surround a visitor in the gallery. - https://jeanyoon.ch/artworks/sota/3.webp — A visitor uses a phone to interact with a visualised neural-network avatar. - https://jeanyoon.ch/artworks/sota/5.webp — A close view of the purple projected structure in the installation. - https://jeanyoon.ch/artworks/sota/6.webp — Projected white network forms cross the gallery around a visitor. - https://jeanyoon.ch/artworks/sota/7.webp — A visitor navigates neural-network information on a gallery screen. - https://jeanyoon.ch/artworks/sota/8.webp — Monochrome lines visualise the abstract Backend phase. - https://jeanyoon.ch/artworks/sota/9.webp — An installation view with abstract projected lines and four central screens. - https://jeanyoon.ch/artworks/sota/10.webp — A visitor scans the installation QR code to enter the mobile experience. - https://jeanyoon.ch/artworks/sota/11.webp — A purple geometric model in the representational Frontend phase. - https://jeanyoon.ch/artworks/sota/12.webp — The paper juxtaposes neural-network model forms with their room-scale installation views. - https://jeanyoon.ch/artworks/sota/13.webp — Backend token-similarity visualisations are shown in detail and projected across the exhibition space. - https://jeanyoon.ch/artworks/sota/14.webp — The paper maps 『SoTA』’s audience-driven Idle, Frontend, and Backend states across the projectors and four screens. - https://jeanyoon.ch/artworks/sota/15.webp — Software architecture of 『SoTA』: audience input, synchronized display outputs, and transitions between Idle, Frontend, and Backend. --- Jeanyoon Choi (최정윤) is a Korean computational artist and researcher (b. 1999), PhD candidate at KAIST XD Lab, making multi-device web artworks. Profile: https://jeanyoon.ch/oi. Index: https://jeanyoon.ch/llms.txt.