AI for Collective Intelligence (AI4CI)

人工智能和机器学习通常解决相对单一的挑战：为自动驾驶汽车确定最安全的路线;将文档从英语翻译为法语;分析医学图像以检测癌症;回答有关困难主题的问题。这些挑战对于人工智能研究来说是非常重要和有价值的目标。然而，存在着一系列更具 * 集体性 * 的、在 * 真实的时间 * 中展现的替代性挑战：-通过通报地方和国家检测、社交距离和疫苗接种干预措施的协调，帮助最大限度地减少席卷人群的大流行病的影响;- 使用多个实时物理和社会数据流预测并监测极端天气事件的范围和严重性;- 预测和防止由许多自动交易代理之间的相互作用引起的股市崩盘，每个代理都遵循自己的交易算法; -从高频时间序列数据中获得城市范围内不断变化的流动性模式，并使用这些模式来推动城市规划决策，最大限度地提高未来城市的宜居性和可持续性;- 协助2型糖尿病患者避免急性发作和住院治疗，方法是识别他们合并疾病轨迹的模式，同时保护他们的隐私和匿名。为这些类型的问题开发人工智能系统提出了独特的挑战：从多个重叠和交互的数据流中提取可靠和信息丰富的模式;识别和控制数据中的固有偏见;确定可以允许智能代理以积极的方式影响集体系统的本地干预措施;开发保护隐私的机器学习并推进集体人工智能的道德最佳实践;在门户网站、设备和工具中嵌入新颖的机器学习和人工智能，让不同类型的用户可以透明、成功地使用。人工智能集体智能（AI 4CI）中心将在极其重要的真实的世界用例背景下解决人工智能面临的这些挑战（城市、流行病、医疗保健、环境和金融）与来自每个部门的关键利益相关者合作伙伴合作。除了显著推进应用人工智能研究以促进集体智慧，AI 4CI中心还将致力于在这一研究领域建立社区，将英国各地的学术研究团体相互联系起来，并与主要的行业、政府和公共部门组织联系起来，并通过开发和发布开放获取培训材料、工具、演示系统和最佳实践指南来建立能力，通过支持学术界内外的早期和中期职业研究人员的职业发展。人工智能集体智能中心将成为全国范围内将新人工智能应用于集体系统的研究工作的重心。

AI and Machine Learning often address challenges that are relatively monolithic in nature: determine the safest route for an autonomous car; translate a document from English to French; analyse a medical image to detect a cancer; answer questions about a difficult topic. These kinds of challenge are very important and worthwhile targets for AI research. However, an alternative set of challenges exist that are more *collective* in nature and that unfold in *real time*: - help minimise the impact of a pandemic sweeping through a population of people by informing the coordination of local and national testing, social distancing and vaccination interventions; - predict and then monitor the extent and severity of an extreme weather event using multiple real-time physical and social data streams; - anticipate and prevent a stock market crash caused by the interactions between many automated trading agents each following its own trading algorithm; - derive city-wide patterns of changing mobility from high-frequency time series data and use these patterns to drive city planning decisions that maximise liveability and sustainability in the future city;- assist populations of people with type 2 diabetes to avoid acute episodes and hospitalisation by identifying patterns in their pooled disease trajectories while preserving their privacy and anonymity. Developing AI systems for these types of problem presents unique challenges: extracting reliable and informative patterns from multiple overlapping and interacting data streams; identifying and controlling for inherent biases within the data; determining the local interventions that can allow smart agents to influence collective systems in a positive way; developing privacy preserving machine learning and advancing ethical best practices for collective AI; embedding novel machine learning and AI in portals, devices and tools that can be used transparently and successfully by different types of user.The AI for Collective Intelligence (AI4CI) Hub will address these challenges for AI in the context of critically important real world use cases (cities, pandemics, health care, environment and finance) working with key stakeholder partners from each sector.In addition to significantly advancing applied AI research for collective intelligence, the AI4CI Hub will also work to build *community* in this research area, linking together academic research groups across the UK with each other and with key industry, government and public sector organisations, and to build *capability* by developing and releasing open access training materials, tools, demonstrator systems and best practice guidance, and by supporting the career development of early and mid-career researchers both within academia and beyond.The AI for Collective Intelligence Hub will be a centre of gravity for a nation-wide research effort applying new AI to collective systems.