Kim Gerdes（凯德金）

Kim Gerdes（凯德金）

Professor of Natural Language Processing, Université Paris-Saclay, France

Personal Profile

Kim Gerdes,obtained a Master's in Mathematics from the University Paris-Sud and ÉcolePolytechnique, a PhD in Linguistics from the University of Paris-Cité.Previously an Associate Professor of Computational Linguistics at SorbonneNouvelle, he now serves as a full Professor of Natural Language Processing atthe University Paris-Saclay, France, where he led the AI track of the computerscience master's program.

Throughouthis career, Kim Gerdes has focused on Natural Language Generation, generativemodels, and syntactic analysis of non-standard language. In 2011, he co-foundedthe international Dependency Linguistics conference series (depling.org) andhas played a pivotal role in leading or co-directing numerous natural languageprocessing projects. These projects resulted in the creation of the firstsyntactically annotated corpora for spoken French, Nigerian Pidgin, spokenMandarin, and Cantonese.

Between2010 and 2013, Gerdes was an invited researcher at the University Pompeu Fabrain Barcelona, the Chinese Academy of Sciences' Institute of Pattern Recognitionin Beijing, and the Xi'an Jiaotong University. He has been the (co-) editor-in-chiefof 11 books, authored 6 book chapters, published 7 academic journal papers, andover 60 international conference papers, with an h-index of 21.

Currently,Kim Gerdes leads technology development at the start-up project Qatent(https://qatent.com), which focuses on developing intelligent software forinvention patent authoring at the intersection of Machine Creativity andsequence-to-sequence models for text generation.

The main theme of the speech

I’m sorry,Dave, I will write the NLP paper myself

The adventof advanced AI systems, especially Large Language Models, might just be pavingthe way for NLP researchers to transition from trailblazers to relics. Theexponential growth in NLP publications each year could soon become a job forour AI creations. This keynote takes a dive into how AI impacts every stage ofan NLP research paper's lifecycle—from conjuring up ideas and verifying theirnovelty to proposing methodologies, executing experiments, and penning thefinal manuscript. Imagine AIs sifting through mountains of research to generatefresh neural network architectures or innovative uses of transfer learning, allwhile ensuring your paper meets the fussy standards of top conferences. Despitethese tech marvels, there still seems to be a need for human touch (for now).We highlight the current research on AI support for NLP scientists, look in particularat the impact on the paper reviewing process, and speculate wildly on theshort-term future where AI might just outshine us all. This session offers acomprehensive yet entertaining look at how AI could reshape the NLP landscape,all while emphasizing that, fortunately, human creativity still holds thereins—for now.

Kim Gerdes（凯德金）

法国巴黎萨克雷大学自然语言处理教授

个人简介：

拥有巴黎南大学、巴黎综合理工学院数学硕士学位，巴黎西岱大学语言学博士学位，曾担任巴黎第十大学及巴黎第三大学计算语言学的副教授。现为法国巴黎萨克雷大学自然语言处理教授，负责人工智能方向计算机科学硕士课程。

凯德金教授一直专注自然语言生成、生成模型以及非标准语言的句法分析。2011年他联合创立了国际依存语言学系列会议international Dependency Linguistics conference series （depling.org），主持或共同主持了多项自然语言处理项目，并构建了法语口语、尼日利亚皮钦语、普通话以及粤语口语的首个句法标注语料库。

2010年至2013年期间，凯德金教授受邀在在多家大学，如巴塞罗那庞培法布拉大学、北京中国科学院模式识别实验室以及西安交通大学担任研究员。主编或联合主编11部著作，撰写6本著作章节，发表7篇学术期刊论文及60多篇国际会议论文，h指数21。目前凯德金教授在初创公司Qatent（https://qatent.com）中负责技术开发，该项目是集机器创造技术、文本生成序列以及序列模型技术于一体，是一项应用于撰写发明专利文档的智能软件。

主旨发言概要：

对不起，Dave (戴夫)，我自己可以完成这篇自然语言处理的论文！

随着高级人工智能系统的兴起，尤其是大型语言模型的出现，自然语言处理（NLP）的研究者正面临着从开拓者走向淘汰者的局面。要知道每年NLP的论文数量呈指数倍增长，论文写作很快便成为AI创作的一项任务。在本次主旨发言中，本人将深入介绍AI是如何影响一篇NLP研究论文“生命周期”的每一阶段：从新点子的构思，到新颖性的论证，再到方法论的提出、实验的实施，到达最后终稿的完成。要知道AI是从海量研究中筛选出了新的神经网络架构，或创新性的迁移学习，而且又可确保相关论文符合顶级会议的严苛标准。尽管有这些神奇的技术，但目前似乎（暂时性）仍然需要人的“触碰”。为此我将重点介绍当前有关AI对NLP科研人员给予支持的相关研究，并特别关注对论文评审过程的影响。本人斗胆预估在短期未来里AI可能会超越我们所有人。最后本人将对AI如何重塑NLP领域的未来，做一个整体但有趣的展望。所幸的是，还需要强调，目前阶段——人类仍掌握着创造力的主权。