Insider Vision: AI for Real-time Learning and Facilitation of Business Communication from Multimodal Media Streams

近年来，人工智能（AI）在人类实践和生活的大量领域取得了爆炸式的发展和成功的部署。能够从视觉和音频记录中检测和识别目标的系统是可用的开源材料。尽管这些系统是免费访问和使用的，但却非常有效。然而，尽管取得了这样的成功，它们的部署并没有远远超出原来的目标检测/识别任务。另一方面，对这种技术的更复杂的使用有很高的需求，例如基于多个异构数据流的自动化高级行为分析。在这个项目中，我们将关注开发一种技术，用于自动实时分析处理共同问题的相互作用的个人群体，随后创建最适合并优化群体互动和决策制定的实践环境和协作机制，以成功地交付任务。随着组织边界的模糊和技术跨越学科和边界，团队的协作能力正在成为一个优先考虑的问题。团队通常缺乏技能和认知工具，使他们能够优化产生想法和协作性质的工作。利用人工智能来推断和可视化实时团队互动机制，为彻底改变人类互动和智能系统如何相互促进提供了机会。该项目将使传统的最先进的人工智能向为共同目标服务的专业人工智能集成系统/社区迈出重要一步。该项目本质上是多学科的，也涉及到一个工业合作伙伴，CALM。开发一种能够跟踪、检测、分析和建模所有交互方的技术，其任务的复杂性是前所未有的。然而，一旦开发出来，这项技术将有能力在任何商业环境中使用，在商业环境中，团队合作对成功至关重要。越来越多的团队合作不再是内部一致的团队在单一组织边界内工作的问题。工作现在是跨界的——组织、文化、专业知识——包括虚拟的和间接的，以及面对面的合作和互动。该项目具有强大的翻译组件，因为该技术可以在广泛的优先领域中部署。特别的例子是安全和医疗保健——对国家具有高度重要性和价值的领域，以及资本密集型合资企业，如大规模基础设施或工程项目，HS2就是一个当前相关的例子。

Recent years have seen explosive development and successful deployment of Artificial Intelligence (AI) across large swathes of Human Practice and life. Systems capable of object detection and recognition from visual and audio recordings are available as open-source material. These systems, despite being free-to-access and use, are remarkably potent. And yet despite this success, their deployment does not go far beyond the original object detection/recognition tasks. On the other hand, there is a high demand for more sophisticated use of this technologies for e.g. automated higher-level behavioural profiling based on multiple heterogenous data streams. In this project, we will be concerned with developing a technology for automated real-time profiling of groups of interacting individuals working on a common problem followed by creating environments and collaborative mechanisms of practice that that are best suited to, and optimise, group interaction and decision making to successfully deliver tasks. As organisational boundaries blur and technology spans disciplines and borders, the collaborative capacity of teams is becoming a priority issue. Teams are often poorly equipped with skills and cognitive tools that enable them to optimise work that is generative of ideas and collaborative in nature. Utilising AI to infer and visualise mechanisms of real-time team interaction offers the opportunity to revolutionise how human interaction and intelligent systems can enable each other. The project will make a major step forward from conventional state-of-the art to integrated systems/communities of specialized AIs serving a common goal. The project is inherently multi-disciplinary and involves an industrial partner, CALM, too. The sheer complexity of the task - to develop a technology capable of tracking, detecting, profiling, and modelling all interacting parties is unprecedented. Yet, when developed the technology will have a capacity to be used in any commercial environment where team working is essential to success. Increasingly team-working is not a question of internally coherent groups working within single organisational boundaries. Work is now cross-boundary - of organisation, culture, expertise - and involves virtual and indirect, as well as face to face collaboration and interaction. The project has strong translational components as the technology can be deployed in a broad range of priority areas. Particular examples are security and healthcare - areas of high national importance and value, as well capital intensive joint-ventures, such as massive infrastructure or engineering programmes, HS2 being a current relevant example.