Periodic Reporting for period 2 - HumanE-AI-Net (HumanE AI Network)
Periodo di rendicontazione: 2022-03-01 al 2023-08-31
There is a strong consensus that AI will beget changes far more profound than any other technological revolution in human history. The HumanE AI Network (hereafter referred to as HumanE AI Net or HumanE AI) aims to leverage the synergies between the involved centers of excellence to develop the scientific foundations and technological breakthroughs needed to shape the AI revolution in a direction that is beneficial to humans both individually and societally, and that adheres to European ethical values and social, cultural, legal, and political norms. The core challenge is the development of robust, trustworthy AI systems capable of what could be described as “understanding” humans, adapting to complex real-world environments, and appropriately interacting in complex social settings. The aim is to facilitate AI systems that enhance human capabilities and empower individuals and society as a whole, while respecting human autonomy and self-determination.
After a project setup phase initiating the internal and external collaboration mechanisms the first 18 months were focused on engaging with the research questions posed in the proposal within WPs 1-5 and conducting a series of concrete high impact activities to connect to the community. With respect to the research work, we have set up a micro project network based on a unique collaboration model that allows agile interaction between partners, interfacing related activities outside the project and easy engagement with researchers outside the consortium. Each of our micro projects is a cooperation of two or more partners over a period of typically 1-4 months aimed at producing a tangible outcome (paper, data set, demo, tutorial etc.) to be made available to the community through the AI4EU platform and other venues, and appropriately promoted in the community. A micro project can (and is encouraged to) include activities that partners have outside the core HumanE AI project as well as researchers from outside the consortium). Nearly 70 micro projects spanning the large majority of the project partners (some are not involved in research and are thus not part of the microprojects) have been initiated resulting in 82 project publications, incl. Nature, PNAS, Phys.Rev Artificial Intelligence etc papers. A major result of this work has been the updated research agenda which includes a novel conceptual framework for human AI collaboration, a notion of shared representations centered around of narratives and the expansion of the definition of Aíitrustworthiness and explainability in terms of human computer interaction (systems that humans (both individually and as a society) feel they understand and are comfortable trusting rather than systems that “only” fulfill certain hard technical specification). We have also identified the development of a specific research methodology for Human Centric AI that combines requirements form AI, HCI and social science as a core challenge and made available first components of such an infrastructure (dedicated data sets).
Presentations of the ideas and the initial results of all those microprojects are accessible through the project website (https://www.humane-ai.eu) which also includes information on all other project activities, the project vision, project plans and multimedia materials we created as part of the dissemination work. The project has also created a presence on the new AI4EU platform https://www.ai4europe.eu/ai-community/projects/humane-ai-net including the placement of the microproject tinformation and results as research bundles https://www.ai4europe.eu/ai-community/projects/humane-ai-net .
A key activity in the first 18 months has been a series of online events most of which were open to a broader community and overall attracted well over 1000 participants worldwide. It started with a moderated (by a professional journalist) panel on the goals of the project, followed by a scientific workshop on interaction between human memory and AI, by a panel on AI regulation (that included participation by Eva Kaili, Member of the European parliament), an AI & Society Roundtable (with such prominent international speakers as Prof. Sandy Pentland from MIT or Prof. Stuart Russell from Berkeley) and a research event on Human and AI Collaboration, Interaction, and Partnership (see https://www.humane-ai.eu/event/ for a full list of events).
As a major contribution to education and outreach, the consortium has organized the Advanced Course on AI (ACAI), a specialized course in Artificial Intelligence sponsored by EurAI in odd-numbered years. In 2021the topic was Human Centered AI and the HumanE AI-Net project did not only organize and host the event, but also provided for more than half of the lectures based on the work and competencies of the project. The content of the lectures is currently being prepared for publication by Springer and all slides will be made available as a blueprint for an entire Human Centric AI course that can be adapted by Universities throughout Europe. This 4-day event in hybrid form included 14 tutorials and 4 keynotes and was attended by 60 students in Berlin and 50 students online. It concluded with an open session (with around 250 participants) with keynotes by Prof. Ben Schneiderman of Maryland University and Turing Award winner Prof. Yoshua Bengio of Montreal. Currently preparation is in full swing for the Dagstuhl Seminar on Human Centric AI to take place as a physical event in June 2022. HumanE AI Net is also an active contributor to the AIDA doctoral academy.
Presentations of the ideas and the initial results of all those microprojects are accessible through the project website (https://www.humane-ai.eu) which also includes information on all other project activities, the project vision, project plans and multimedia materials we created as part of the dissemination work. The project has also created a presence on the new AI4EU platform https://www.ai4europe.eu/ai-community/projects/humane-ai-net including the placement of the microproject tinformation and results as research bundles https://www.ai4europe.eu/ai-community/projects/humane-ai-net .
A key activity in the first 18 months has been a series of online events most of which were open to a broader community and overall attracted well over 1000 participants worldwide. It started with a moderated (by a professional journalist) panel on the goals of the project, followed by a scientific workshop on interaction between human memory and AI, by a panel on AI regulation (that included participation by Eva Kaili, Member of the European parliament), an AI & Society Roundtable (with such prominent international speakers as Prof. Sandy Pentland from MIT or Prof. Stuart Russell from Berkeley) and a research event on Human and AI Collaboration, Interaction, and Partnership (see https://www.humane-ai.eu/event/ for a full list of events).
As a major contribution to education and outreach, the consortium has organized the Advanced Course on AI (ACAI), a specialized course in Artificial Intelligence sponsored by EurAI in odd-numbered years. In 2021the topic was Human Centered AI and the HumanE AI-Net project did not only organize and host the event, but also provided for more than half of the lectures based on the work and competencies of the project. The content of the lectures is currently being prepared for publication by Springer and all slides will be made available as a blueprint for an entire Human Centric AI course that can be adapted by Universities throughout Europe. This 4-day event in hybrid form included 14 tutorials and 4 keynotes and was attended by 60 students in Berlin and 50 students online. It concluded with an open session (with around 250 participants) with keynotes by Prof. Ben Schneiderman of Maryland University and Turing Award winner Prof. Yoshua Bengio of Montreal. Currently preparation is in full swing for the Dagstuhl Seminar on Human Centric AI to take place as a physical event in June 2022. HumanE AI Net is also an active contributor to the AIDA doctoral academy.
The project has been very active in innovation activities set up including a series of events meant at bringing academic innovators together with relevant industrial and financial players and fostering the translation of research results into potential product ideas. This included the AI Idea Price we connect students with corporate decision makers, special awards and its own HumaneAI Net jury at the Cashwalk event that connects European startups looking for investors with the international VC scene and networking events with SME focus like "Best Practice Digitization and AI for the SMEs”.
Finally, the project has been actively interacting with the other ICT48 initiatives participating in a variety of events. At the international level, Humane-AI-Net has been shortlisted in the 30 most promising initiatives in Responsible AI by the GPAI, the Global Partnership on Artificial Intelligence http://gpai.ai a global multi-stakeholder initiative which aims to bridge the gap between theory and practice on AI by supporting cutting-edge research and applied activities on AI-related priorities. GPAI involves several Humane-AI researchers as experts.
Finally, the project has been actively interacting with the other ICT48 initiatives participating in a variety of events. At the international level, Humane-AI-Net has been shortlisted in the 30 most promising initiatives in Responsible AI by the GPAI, the Global Partnership on Artificial Intelligence http://gpai.ai a global multi-stakeholder initiative which aims to bridge the gap between theory and practice on AI by supporting cutting-edge research and applied activities on AI-related priorities. GPAI involves several Humane-AI researchers as experts.