AccelNet-Design: Accelerating Research Collaboration on Artificial and Natural Intelligences (ARCANI)

研究思维和认知的领域正在蓬勃发展。认知科学已经蓬勃发展。我们对动物思维的理解已经从根本上扩展了。人工智能研究已经突破了长期存在的里程碑，比如下围棋或写可信的文章。这些领域以不同的形式研究相同的现象--智能--包括自然的和人工的。但它们仍然是孤立的，每个都缺乏连贯的基础。这在人工智能领域最为明显，我们设计强大人工智能系统的能力远远超过了我们理解它们的能力。通过在研究自然和人工智能的主要学科之间建立对话，我们的项目将为加快人工和自然智能研究合作(ARCANI)制定路线图。为了创建这一路线图，ARCANI将连接几个国际研究网络，利用跨学科、职业阶段和国家的见解和对话。这些研究网络的整合还将有助于多样化地参与自然和人工智能的研究。它将培训一批早期职业研究人员以新的方式开展科学和国际合作。通过现有的多元化智能暑期研究所，ARCANI团队在促进来自代表性不足的群体和发展中国家的研究人员之间的合作和领导力方面有着良好的记录。ARCANI使我们能够扩大我们的努力，改善许多学科的年轻学者的培训、合作机会和技能发展。ARCANI路线图将导致从动物行为学到人工智能的理论基础的发展。它将建立一个充满活力的交易区，共享新的方法、数据和计算框架。它将把基本概念和经验见解转化为实用技术。这些跨学科的努力将改变人类对环境的看法，与计算机系统互动，并建立人工智能，以支持更广泛的科学、艺术和人类努力。这些成果将加强美国在许多关键研究领域的科学领导地位。我们的网络网络将来自生物、认知、社会和计算机科学的研究人员聚集在一起，开发新的方法来加速人工和自然智能研究合作(ARCANI)。ARCANI的目标是打破学科孤岛，开发一个探索人类、动物和机器智能的全面框架。我们将把行为学、复杂性科学、比较认知、神经科学、机器人和人工智能领域的网络与每个宜居大陆的节点联系起来。我们已经确定了通过网络到网络协作加速发展的三个有前途的领域：(1)智能的理论基础(定义)；(2)利用数据革命重新设计认知研究(设计)；以及(3)人工智能和机器人的自然主义方法(部署)。每个领域都将由一个跨网络工作组进行探索，举行几次虚拟会议和一次面对面会议。早期职业研究人员(ECR，包括研究生和博士后)将发挥重要作用；多元化智能暑期学院(DISI)将帮助招聘多样化和极具天赋的ECR。工作组将讨论三个问题：现有知识的差距是什么？在他们的网络中，谁最有能力走到一起解决这些问题？AccelNet如何支持可持续的国际合作，以缩小这些差距？这些问题将推动对现有研究的调查、关键概念的定义和对障碍的坦率评估。工作组将在白皮书中总结其调查结果，我们将把这些白皮书作为议程设置文件公布。工作组成员将与领导团队一起起草2024年实施提案，为在人工智能和自然智能研究的基础和应用方面实现跨网络合作提供路线图。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The fields that study mind and cognition are flourishing. Cognitive science has blossomed. Our understanding of animal minds has radically expanded. AI research has blazed past longstanding milestones like playing Go or writing credible essays. These fields study the same phenomenon—intelligence—in diverse forms, both natural and artificial. But they remain siloed, and each lacks a coherent foundation. This is most glaring in AI, where our capacity to engineer powerful AI systems has vastly outpaced our capacity to understand them. By building dialogue between the major disciplines studying natural and artificial intelligences, our project will produce a roadmap for Accelerating Research Collaboration on Artificial and Natural Intelligences (ARCANI). To create this roadmap, ARCANI will link several international research networks, drawing on insights and conversations that cross disciplines, career stages, and countries. The integration of these research networks will also help diversify participation in the study of natural and artificial intelligences. It will train a cohort of early career researchers in new ways of doing science and international collaboration. Through the existing Diverse Intelligences Summer Institute, the ARCANI team has an excellent record of fostering collaboration and leadership amongst researchers from underrepresented groups and developing nations. ARCANI allows us to scale up our efforts, improving training, collaboration opportunities, and skill development for young scholars across many disciplines. The ARCANI roadmap will lead to the development of theoretical foundations spanning from animal behavior to AI. It will establish a vibrant trading zone for sharing novel methods, data, and computational frameworks. And it will translate foundational concepts and empirical insights into practical technologies. These interdisciplinary efforts are poised to transform the ways humans think about the environment, interact with computer systems, and build AI to support science, art, and human endeavors more broadly. Such outcomes will enhance US scientific leadership in many critical lines of research.Our Network of Networks brings together researchers from the biological, cognitive, social, and computer sciences to develop new approaches to Accelerating Research Collaboration on Artificial and Natural Intelligences (ARCANI). ARCANI aims to break down disciplinary silos and develop a comprehensive framework for exploring human, animal, and machine intelligences. We will link networks in ethology, complexity science, comparative cognition, neuroscience, robotics, and AI, with nodes in every habitable continent. We have identified three promising areas for accelerated advance through network-to-network collaboration: (1) theoretical foundations of intelligence (DEFINE); (2) leveraging the data revolution to redesign the study of cognition (DESIGN); and (3) naturalistic approaches to AI and robotics (DEPLOY). Each area will be explored by a cross-network working group, holding several virtual meetings and one in-person meeting. Early career researchers (ECRs, including grad students and postdocs) will play a prominent role; the Diverse Intelligences Summer Institute (DISI) will help recruit diverse and exceptionally gifted ECRs. Working groups will address three questions: WHAT are the gaps in existing knowledge? WHO in their networks is best positioned to come together to address them? HOW can AccelNet support sustainable international collaboration to close those gaps? These questions will drive surveys of existing research, definition of key concepts, and frank assessment of barriers. Working groups will summarize their findings in white papers, which we will publish as agenda-setting documents. Working group members will join the leadership team in crafting a 2024 Implementation proposal, providing a roadmap toward transformational cross-network collaboration on the foundations and applications of research on artificial and natural intelligences.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.