-------- Forwarded Message -------- Subject: [WI] [meetings] 2nd CFP: New Foundations for Human-Centered AI (an ECAI workshop) Date: Sun, 9 Feb 2020 23:54:14 +0100 From: Alessandro Saffiotti asaffio@aass.oru.se Reply-To: Alessandro Saffiotti asaffio@aass.oru.se To: wi@lists.kit.edu
====================================================================== = NeHuAI-2020 = 1st International Workshop on New Foundations for Human-Centered AI = June 9, Santiago de Compostela, Spain (Part of ECAI-2020) = http://nehuai2020.aass.oru.se/ ======================================================================
SYNOPSIS --------
In June 2018, the European Commission has appointed a "AI High Level Expert Group" (AI-HLEG) to support the implementation of the European Strategy on Artificial Intelligence. One of the first results of the AI-HLEG has been to deliver ethics guidelines on Artificial Intelligence (https://ec.europa.eu/futurium/en/ai-alliance-consultation/guidelines). These guidelines put forward a human-centered approach to AI, and list seven key requirements that human-centered, trustworthy AI systems should meet, summarized by the following headers:
1. Human agency and oversight 2. Technical robustness and safety 3. Privacy and data governance 4. Transparency 5. Diversity, non-discrimination and fairness 6. Societal and environmental wellbeing 7. Accountability
Many of today's most popular AI methods, however, fail to meet these guidelines: making them compliant is a scientific endeavor that is as crucial as it is challenging and stimulating. Systems based on deep learning are a case in point: while these systems often provide impressive results, their ability to _explain_ these results to the user is very limited, challenging requirements 4 and 7; in most cases we lack ways to formally _verify_ their correctness and assess their boundary conditions, challenging requirement 2; and we don't yet have methods to allow humans to _collaboratively_ influence or question their decisions, challenging requirement 1. Similar criticalities are present in many other popular AI methods.
This full day workshop will collectively address the fundamental questions of what are the scientific and technological gaps that we have to fill in order to make AI systems _human-centered_ in terms of the above guidelines.
PAPER SUBMISSION ----------------
Contributions are seeked on new foundations for building Human-Centered AI systems, able to comply with AI-HLEG recommendations. Contributions are seeked in the form of full papers (max 7 pages plus references) presenting mature results, or position papers and reports of relevant ongoing work (max 4 pages including references). More specific topics include, but are not limited to:
- Explainable AI - Verifiable AI - Technical robustness and safety of AI systems - Collaboration between humans and AI systems - Integrating model-based and data-driven AI - Integrating symbolic- and sub-symbolic AI - Mixed initiative AI-Human systems - Proactive AI systems in human environments - Understanding and naturally interacting with humans - Understanding and interaction in complex social settings - Reflexivity and expectation managament - Integrating Learning, Reasoning and Acting in AI systems
Papers should be formatted according to the ECAI2020 formatting style, available at the ECAI2020 website (ecai2020.eu). Submissions are not anonymous.
Submit your paper by February 25 via Easychair here:
https://easychair.org/conferences/?conf=nehuai2020
ORGANIZERS ----------
- Alessandro Saffiotti (Orebro University, Sweden) - Luciano Serafini (Fondazione Bruno Kessler, Trento, Italy) - Paul Lukowicz (DFKI, Kaiserslautern, Germany)
This worskhop is jointly organized by AI4EU (ai4eu.eu), the EU landmark project to develop a European AI on-demand platform and ecosystem; and by Humane-AI (humane-ai.eu), the EU FET preparatory action devoted to designing a European research agenda for Human Centered AI.
MORE INFORMATION ----------------
http://nehuai2020.aass.oru.se/
======================================================================
-- Mailing-Liste: wi@lists.kit.edu Administrator: wi-request@lists.kit.edu Konfiguration: https://www.lists.kit.edu/wws/info/wi