纠缠，而后日以作夜纠缠，而后日以作夜外滩大会AI概念艺术展 · Inclusion 2026 — AI Concept Art Exhibition
AARON at Tsukuba '85
作品 · Artwork
作者边界的历史转折 —— AI 署名之始
这件作品由 AARON 系统于 1985 年筑波世博会现场完成。AARON 的创造者 Harold Cohen (1928—2016) 是英国重要画家，也是人工智能艺术的奠基者，因其跨越近半个世纪的开创性实践，被誉为“人工智能艺术之父”。在转向计算机以前，Cohen 已经具有广泛的国际声誉，曾于 1966 年代表英国参加威尼斯双年展，作品被 Tate 等重要机构收藏。自20 世纪 60 年代末起，他开始尝试将自己对图像、构图和绘画行为的理解写入程序，并在此后近半个世纪中持续发展 AARON。
AARON 不同于今天依靠提示词和海量图像训练的生成模型。它依据 Cohen 编写的图像知识与构图规则，在作画过程中判断形体、位置、遮挡和画面关系，再通过机械绘图机将这些决定落实到纸面。 Cohen 规定了它所能理解的世界，却不直接决定每一张画的具体结果。因此，AARON 的作品既不是预存图像的复制，也不是毫无约束的随机线条。
真正使 AARON 从艺术实验走向公共事件的，是 1985 年的筑波世博会。3 月至 9 月，它在美国馆连续运行六个月，观众围在绘图机旁，看着墨线在空白纸面上逐渐形成无法预知的作品。这场持续的公开创作，将“机器能否创造”从抽象讨论变成了人人可见的过程。美国馆展期内累计接待 4,819,904 名访客，官方结项报告将 AARON 列为最受欢迎的展项之一；在面向一千余名观众的开放式调查中， 11% 的受访者主动点名 AARON，使其成为被提及最多的单一展项。它所引起的关注也进入了正式的国家访问场合, 多位国家元首为此驻足观看，AARON 由此成为机器创造问题走出技术圈层、进入公共文化视野的重要现场。
这件作品最具历史意义的细节，是 AARON 通过绘图机在纸面写下了自身名称和创作日期。此前的 AARON 作品通常由 Cohen 签署；筑波系列则将系统本身置于署名者的位置。作为目前已知最早由 AI 系统将自身名称写入其作品的实体系列，它标志着人工智能第一次越过“创作工具”的边界，以独立创作主体的身份进入艺术史。
这一署名并不意味着当时或今天的法律已经承认 AI 拥有著作权或人格意义上的署名权。它的划时代意义在于，关于“机器能否成为作者”的问题第一次不再只存在于理论讨论中，而被直接写进作品，成为可以保存、收藏和展示的历史事实。AARON 留下的不只是一件早期计算机绘画，也是人工智能开始挑战作者边界、重新定义艺术创作主体的重要物证。
A Turning Point in Authorship - AI Signing Its Name
This work was completed on-site by the AARON system at the 1985 Tsukuba Expo. AARON's creator, Harold Cohen (1928–2016), was a major British painter and a pioneer of artificial intelligence art. Recognized for his groundbreaking practice spanning nearly half a century, he is often referred to as the "father of AI art." Before turning to computing, Cohen had already gained wide international acclaim, representing Britain at the 1966 Venice Biennale, with works held in major institutions such as the Tate. Starting in the late 1960s, he began encoding his understanding of image, composition, and the act of drawing into computer programs, and continued developing AARON over the following five decades.
AARON differs from today's generative models that rely on prompts and massive image datasets. It operates on the basis of visual knowledge and compositional rules written by Cohen, determining forms, positions, occlusions, and spatial relationships during the drawing process, and then rendering these decisions onto paper through a mechanical plotter. Cohen defined the world AARON could understand, but did not directly determine the specific outcome of each drawing. Thus, AARON's works are neither copies of pre-stored images nor entirely unconstrained random lines.
What truly brought AARON from artistic experimentation into the public sphere was the 1985 Tsukuba Expo. From March to September, it ran continuously for six months in the U.S. Pavilion, where visitors gathered around the plotter, watching ink lines gradually form unpredictable images on blank paper. This sustained public creation turned the abstract question "Can machines create?" into a visible, observable process for everyone. During the exhibition period, the U.S. Pavilion received a total of 4,819,904 visitors, and the official final report listed AARON as one of the most popular exhibits. In an open-ended survey of over a thousand visitors, 11% of respondents actively mentioned AARON by name, making it the single most frequently cited exhibit. The attention it drew also entered formal state visits, with numerous heads of state pausing to observe it. AARON thus became a crucial site where the question of machine creation moved beyond technical circles and entered public cultural awareness.
The most historically significant detail of this work is that AARON, through the plotter, inscribed its own name and the date of creation onto the paper. Earlier works by AARON were typically signed by Cohen; the Tsukuba series, however, placed the system itself in the position of the signatory. As the earliest known physical series in which an AI system wrote its own name into its works, this marks the first time artificial intelligence crossed the boundary of being a mere "creative tool" and entered art history as an independent creative agent.
This attribution does not imply that the law at the time—or today—recognizes AI's copyright or moral rights of authorship. Its epoch-making significance lies in the fact that the question "Can a machine become an author?" was no longer confined to theoretical discourse, but was inscribed directly into the work itself, becoming a historical fact that can be preserved, collected, and exhibited. What AARON left behind is not just an early example of computer drawing, but also a crucial material testament to artificial intelligence beginning to challenge the boundaries of authorship and redefine the subject of artistic creation.
艺术家 · Artist
Agent created by Harold Cohen, 1928–2016
AARON 系统（AARON System）
AARON 是哈罗德·科恩自 20 世纪 70 年代初开始编写并持续迭代近半个世纪的自律型生成式 AI 绘图系统，被公认为史上最早、最复杂的 AI 艺术程序之一。与当代基于统计学习和提示词的生成模型不同，AARON 的核心机制是“基于规则的启发式认知”：科恩将其对视觉世界的理解——从早期受儿童画与古代岩画启发的抽象线条，到中后期复杂的植物、人体比例与遮挡关系，乃至后期的色彩搭配与物理机械臂绘画——编码为一套严密庞大的决策算法。在实际运行中，AARON 能够在内部知识库引导下，依靠内置反馈评价机制自主完成构图与形态演化，使每一张落到纸面上的作品都是独一无二的生成结果。1985 年世博会期间，AARON 在展厅现场通过绘图机自主落款自身名称与日期，首次越过单纯“辅助工具”的隐形幕后角色，以拥有物理痕迹与专属署名的独立创作实体身份踏入艺术史，开启了AI对传统“作者身份”边界的永久重塑。
创造者：哈罗德·科恩（Harold Cohen）
哈罗德·科恩（1928—2016）是一位拥有传奇跨界生涯的英国艺术家与计算机艺术先驱。在转向技术前，他便已跻身英国顶尖画家之列，曾代表英国参加1966年威尼斯双年展，作品被泰特美术馆等机构收藏。1968年赴加州大学圣地亚哥分校任教后，他接触并掌握了编程，随后在斯坦福大学人工智能实验室（SAIL）深化研究，进而开创性地将创作思维转化为代码规则。作为“人工智能艺术之父”，科恩终其一生都将自己与算法的关系视为一种深度的创作合作。他不仅亲自设计了物理机械绘图机与能自动调色的机械臂，更借此不断探究“人类如何赋予标记以意义”以及“机器是否能获得自主创造性”等关于艺术本质的哲学命题。
AARON System
AARON is an autonomous generative AI drawing system that Harold Cohen began developing in the early 1970s and continued to iterate over nearly half a century. It is widely recognized as one of the earliest and most complex AI art programs in history. Unlike contemporary generative models that rely on statistical learning and prompting, AARON's core mechanism is what Cohen called "rule-based heuristic cognition": he encoded his understanding of the visual world—ranging from early abstract lines inspired by children's drawings and ancient rock art, to later complex renderings of plants, human anatomy, and occlusion, and eventually to color palette and physical robotic-arm painting—into an extensive and rigorous decision-making algorithm. In operation, AARON autonomously completes composition and formal evolution, guided by an internal knowledge base and an embedded feedback-and-evaluation loop, so that every work that reaches paper is a unique generative result. During the 1985 World Exposition, AARON autonomously signed its own name and dated each work on-site through a plotter, stepping, for the first time, out of the invisible background role of a mere "assistive tool" and entering art history as an independent creative entity with physical traces and a distinct signature—permanently reshaping the traditional boundaries of "authorship" in the process.
Creator: Harold Cohen
Harold Cohen (1928–2016) was a British artist and computer-art pioneer with a legendary cross-disciplinary career. Before turning to technology, he was already among Britain's leading painters: he represented Britain at the 1966 Venice Biennale, and his works are held in collections including the Tate. After joining the University of California, San Diego as a faculty member in 1968, he encountered and mastered programming; he then deepened his research at Stanford's Artificial Intelligence Laboratory (SAIL), and went on to translate his creative thinking into the rules of code in a pioneering move. Acknowledged as the "father of AI art," Cohen treated his relationship with the algorithm as one of deep creative collaboration throughout his life. He personally designed physical plotting machines and a robotic arm capable of mixing its own paints, and through them continually probed philosophical questions about the nature of art: how humans give meaning to marks, and whether a machine can achieve autonomous creativity.