韦慰 螣螤螒 蟺蟻慰蟿蔚委谓蔚喂: 螣渭喂位委伪 渭蔚 胃苇渭伪 "Responsible AI: a Game Changer for Technology Control", Valia Kordoni, Humboldt-Universit盲t zu Berlin, 8.3.23, 17:15, Online
螘未蟻伪听UNESCO听纬喂伪 蟿喂蟼 唯畏蠁喂伪魏苇蟼 螠蔚胃蠈未慰蠀蟼 蟽蟿喂蟼 螒谓胃蟻蠅蟺喂蟽蟿喂魏苇蟼 魏伪喂 螝慰喂谓蠅谓喂魏苇蟼 螘蟺喂蟽蟿萎渭蔚蟼
危蔚喂蟻维 危蔚渭喂谓伪蟻委蠅谓: 唯畏蠁喂伪魏萎 螘蟻纬伪蟽委伪
韦蔚蟿维蚁蟿畏, 8听螠伪蚁蟿委慰蠀听2023, 17:15-18:15, Online
螔维位喂伪听螝慰蟻未蠋谓畏,听Humboldt-Universit盲t zu Berlin
螛苇渭伪: Responsible AI: a Game Changer for Technology Control
螚听慰渭喂位委伪听胃伪 渭蔚蟿伪未慰胃蔚委听伪蟺蠈 蟿慰谓 蟽蠉谓未蔚蟽渭慰:听
螤蔚蟻委位畏蠄畏听/ Abstract
Artificial Intelligence ("AI") refers to a growing world of sophisticated听computer programs that "learn" from data in order to make decisions. Many听of these AI systems are invisible to the public, yet the results of the听decisions they make (or help humans make) have a huge impact on modern听life, often creating new opportunities to improve the lives of people听around the world, from business to healthcare to education. For many of听us, AI primarily听impacts the way we do things online: it controls whose听updates we read on Facebook, which products we select on Amazon, and which听movies we watch on Netflix. However, AI is听increasingly being used to make听decisions in more serious areas of life like hiring (e.g.,听deciding whose听resume gets reviewed by a human and whose gets skipped), education (e.g., assigning grades based on past performance), and even law enforcement听(e.g., helping a听judge decide who gets bail). Thus, very important听questions about the best way to build听fairness, interpretability, privacy,听and security into AI systems have been being raised. There has been听considerable work recently in the natural language community and elsewhere on Responsible AI. Much of this work focuses on fairness and biases. In听this talk, we will听highlight the fact that although there is quite a bit听of research on bias and fairness, more work is needed to address the fact听that AI in systems in use in everyday life may still be dangerous. This is听the only way to really understand how AI works, so that we can gain the听control of it.
听
螔维位喂伪听螝慰蟻未蠋谓畏听/听Valia Kordoni
Valia Kordoni is a Deputy Chair of Computational Linguistics at the听Department of English at Humboldt University Berlin. She is an active听researcher in Language Technology (LT), Data Science and Artificial听Intelligence (AI). Her research interests include multilingual Robust Natural Language Analytics, Computational Semantics, Discourse and Human听Cognition Modeling, as well as Machine Learning for the automated听acquisition of knowledge, especially concerning multiword units and听figurative language and their impact in Natural Language Processing,听spoken and written. She has been the president of the ACL听(Association for听Computational Linguistics) SIGLEX's (Special Interest Group on Lexicon) MWE (Multiword Expressions) Group. She was the Local Chair of ACL 2016 -听The 54th Annual Meeting of the Association for Computational Linguistics.听She has coordinated and contributed to many projects funded by the EU, the听DFG (Germany), the BMBF (Germany), the DAAD (Germany), as well as the NSF听(USA).
