Artificial Intelligence and the Useful Art Museum: A Cross-Disciplinary Approach Towards Machine Learning and its Implications in the Museum Sphere

本研究将探讨人工智能（AI）对公共艺术博物馆的贡献。通过与艺术和非艺术领域的行业合作伙伴进行基于实践的合作研究，该项目将加深对人工智能在文化环境中的作用的认识，并了解人工智能对公众信任的影响。具体而言，本研究将探讨以下关键问题：(1)人工智能作为策展策略的作用和潜在用途是什么？(2)如何使用人工智能对现有藏品进行解释和分类，并为收购提供信息？(3)在博物馆中使用人工智能将在哪些方面挑战和/或增强公众的信任？这项研究将探索人工智能如何管理、分类和集群大数据集，同时作为博物馆的共同生产者，并将发现嵌入博物馆的新形式的智能的影响；询问算法输出是否与（新的）策展策略、博物馆利益相关者和文化政策相一致。这项研究将质疑机器学习如何为策展实践提供信息，以及它是否会引入偏见或目前无法预测和未知的模式。其目的是推动艺术史话语超越共同的界限，在算法的帮助下收集知识，并在物品、它们的意义和它们在博物馆藏品中的位置之间建立新的联系。在数字人文学科中，“挑战和改变特定的制度结构”尤为重要（Bassett等人，2017），特别是因为博物馆经常被视为社会不平等“构成、复制和加强”的机构（Sandell, 2005）。该项目将对数字人文学科领域做出重大贡献，批判性地反思和理解“这些技术如何运作以构建周围的世界，并以此改变人文学科的知识和实践”（Berry和Fagerjord， 2017）。此外，本研究将探讨人工智能如何帮助培养对公众有用的博物馆的社会使命——从一个“学科博物馆”（Hooper-Greenhill, 1992）转向一个多元化的博物馆，它与数字相适应，并意识到自己的社会责任——成为一个透明的（Rader等人，2018）和有用的地方，观众/用户可以熟悉人工智能，使学者能够研究互动并提供解释。我建议进行基于实践的跨学科和应用研究，通过对策展实践、展览设计和展示的调查来探索研究问题，这些研究利用了人工智能、机器学习和算法的特性，以及观众对人工智能技术（共同）生产、过滤、中介或分类的艺术的反应。该方法将涉及与博物馆以及创意产业和其他部门的合作者的合作，这些合作者在其研究实践中与人工智能合作，并寻求公众参与和合作生产的机会，以测试这些想法。已经同意加入该项目的合作伙伴是曼彻斯特城市画廊和惠特沃斯馆长阿利斯泰尔·哈德森（Alistair Hudson），以及UoM道尔顿核研究所核能系统BNFL主席理查德·泰勒（Richard Taylor）教授，该研究所应用人工智能技术支持核科学研究，并通过其BEAM网络促进跨学科研究。具体而言，该项目将发现实施沉浸式和人工智能技术的新方法，以一种对四个主要研究利益相关者群体有用的方式：博物馆部门/非艺术和文化产业部门/学术界/公众。曼彻斯特大学道尔顿核研究所核能系统BNFL主席，该研究所应用人工智能技术支持核科学研究，并通过其BEAM网络促进跨学科研究。

This proposed research will explore the contribution of Artificial Intelligence (AI) to the public art museum. Through practice-based, collaborative research with industry partners in the arts and non-arts sectors, this project will develop knowledge of the role of AI in a cultural environment and understand what impact AI will have on public trust. Specifically, this research will investigate the following key questions:(1) What is the role and potential uses of AI as a curatorial strategy?(2) How can AI be used to interpret and classify existing collections and inform acquisition?(3) In what ways will the use of AI in museums challenge and/or enhance public trust?This research will explore how AI will curate, classify and cluster big data sets, whilst acting as a co-producer of museums, and will discover the implications of new forms of intelligence embedded within museums; asking whether algorithmic outputs are aligned with (new) curatorial strategies, museum stakeholders and cultural policies. This research will question how ML may inform curatorial practice and whether it will introduce bias or as yet unpredictable and currently unknown patterns. This aims to push the art historical discourse beyond common boundaries, gathering knowledge with the help of algorithms and creating new connections between objects, their meanings and their place within museum collections. It is particularly important in the digital humanities 'to contest and transform particular institutional structures' (Bassett et al., 2017), especially as museums have often been seen as institutions where social inequalities have been 'constituted, reproduced and reinforced' (Sandell, 2005). This project will significantly contribute to the field of digital humanities, to critically reflect and understand 'how these technologies operate to structure the world around them, and in doing so transform humanities knowledge and practice' (Berry and Fagerjord, 2017). Furthermore, this research will explore how AI can help to foster the social mission of useful museums for the public - away from a 'disciplinary museum' (Hooper-Greenhill, 1992) towards a diverse museum that is digitally fit and aware of its social responsibilities - being a transparent (Rader et al., 2018) and useful place where audiences/users can gain familiarity with AI, enabling scholars to research interactions and to provide explanations.I propose to undertake practice-based, interdisciplinary and applied research which will explore the research questions through investigation of curatorial practices, exhibition design and display which draw on AI and the properties of ML and algorithms, and of audience responses to art which is (co)-produced with, filtered, mediated or classified by AI technologies.The methodology will involve partnership with museums and with collaborators from the creative industries and other sectors who are working with AI within their research practices, and who are seeking opportunities for public engagement and co-production in which to test these ideas. Partners who have already agreed to join this project are Alistair Hudson, Director of the Manchester City Galleries and the Whitworth, and Prof Richard Taylor, BNFL Chair in Nuclear Energy Systems at the UoM's Dalton Nuclear Institute, which applies AI technologies to support research in nuclear science and is fostering cross-disciplinary research via its BEAM network. Specifically, this project will discover new ways of implementing immersive and AI technologies in a way that will be useful to four main research stakeholder constituencies:Museum sector/non-arts and cultural industrial sectors/academic/the public.BNFL Chair in Nuclear Energy Systems at the University of Manchester's Dalton NuclearInstitute, which applies AI technologies to support research in nuclear science and is fosteringcross-disciplinary research via its BEAM network.