How Natural Language Processing contributes to society: an interview with Professor Lonneke van der Plas

Professor Lonneke van der Plas has recently assumed the Daccò Chair in Natural Language Processing at the Institute of Argumentation, Linguistics, and Semiotics (IALS) within the Faculty of Communication, Culture, and Society at Università della Svizzera italiana (USI). Additionally, she serves as an adjunct professor at the Faculty of Informatics. On 18 February 2025, she will deliver an inaugural lecture titled "Breaking Barriers with NLP." 

Professor van der Plas, could you provide a brief summary of your educational and professional background, explaining how you became interested in NLP? 

"I have always had an interest in understanding people—their behaviour and actions—and in tackling complex problems. This combination has sparked my fascination with both linguistics and computer science. Linguistics helps explain how people communicate by studying the intricate processes behind human language, while computer science enables us to comprehend various phenomena by analysing large amounts of data and creating abstractions of complex phenomena. I completed the Master in Computer Speech and Language Processing at the University of Cambridge in 2000, a time when the subject was still quite new. After earning my degree, I worked at a machine translation company for less than a year. I had anticipated this work would be exciting (more exciting than staying on to do a PhD). Still, in reality, it was mainly focused on writing translation rules for individual words. I soon started longing to return to research and decided to do a PhD. What fascinated me was to train machines to identify patterns in language, enabling them to solve tasks in a more flexible way. In my PhD, I focused on methods that allow machines to 'understand' the meanings of words by analysing how words are used in texts. These models were the precursors of the large language models underlying currently popular conversational AI tools."

How would you explain the NLP approach to a layman? 

"NLP traditionally combines linguistics and computer science to process, analyse, and generate human language. It is a subfield of Artificial Intelligence that enables computers to do all kinds of language-based tasks. For example, they can converse using fluent language that mimics interaction with another person. Technologies like ChatGPT allow users to communicate with artificial intelligence (AI) systems using natural language instead of just keywords, simulating a real conversation. Many types of AI rely on language, as do search engines, such as Google. For example, if a user types "bike," the system must understand that it could refer to either a bicycle or a motorbike, demonstrating the need for a level of intelligence in these systems". 

What are the most significant contributions this discipline can make to today's society? 

"There are numerous benefits to natural language processing (NLP) applications. Many tools we use daily, such as Alexa, DeepL, and ChatGPT, rely on these technologies. As a result, accessing information and communicating with people in different languages has become much easier. 

NLP can make huge contributions in the medical domain, where there is an abundance of data and clear needs for improvement that have a huge impact on people's lives. During my PhD, we built a medical question-answering system where a user could ask all kinds of questions regarding their health concerns. I am currently involved in a project funded by the SPIRIT programme of the Swiss National Science Foundation (SNSF). We collaborate with a hospital in Mexico, where patients in need of psychological treatment often find themselves on a long waiting list. To address this issue, we will develop a bot designed to identify those who are most urgently in need of psychological support by asking a series of questions prepared by medical specialists. 

I believe there is an important role to play for NLP experts, and that is to help people gain a better understanding of what NLP is and how it works. Many people use NLP tools or are exposed to AI-generated content, yet few are aware of their potential risks. By educating users about NLP-derived tools and their functionalities, we empower them to use these technologies more effectively and to recognise the associated risks, preventing them from becoming overly trusting or overly optimistic. This was a central theme in the seminar on AI and democracy that I recently organised". 

What courses will you teach at USI next semester, and how will they be helpful for the next generation of communicators? 

"During the spring semester, I will teach two courses: Natural Language Processing for Business and Finance and Natural Language Processing for Business and Finance Project. The first will cover the basics of the subject, while the second course will use a project format to offer the opportunity to design NLP-based solutions to real problems from business scenarios while working in interdisciplinary teams. 

The automatic analysis and generation of (unstructured) textual data are being rapidly adopted in the corporate world across departments, notably in communication and finance. Most communication professionals now use LLM-powered AI for content generation. Also, in finance, NLP powers the use of textual data for research and market intelligence and contributes to the development of FinTech products and services. For future professionals, it has become critical to understand how such NLP methods work, how they can help accelerate analysis and content creation, and where their limitations and risks lie. 

Another goal of these courses is to form interdisciplinary teams that include students from the Faculty of Communication, Culture and Society, the Faculty of Economics, and the Faculty of Informatics. In these teams, each student will be able to find a role that suits them best, collaborating with individuals with different skills and expertise. I believe that working with individuals from various fields of study is highly valuable. It allows us to appreciate the unique contributions that each discipline can make, highlighting that everyone has a role to play."

Why did you choose to come to USI? What environment did you find? 

"What attracted me to USI was the possibility of conducting interdisciplinary work, which is fundamental to NLP. Fondazione Aldo e Cele Daccò created a very interesting position in this respect, which made it possible to start building an interdisciplinary group with available funding. My role allows me to connect two faculties: the Faculty of Communication, Culture and Society and the Faculty of Informatics, where I will also be teaching courses in the upcoming academic year and supervising students. I truly enjoy acting as a bridge between these different disciplines, which have much to offer one another. At USI, I have found a very lively research environment, with many opportunities for collaboration also outside the two faculties; in short, it is a very enriching environment. I wish I had two of me to manage all the projects I've been involved in during recent months! 

Another interesting aspect is the unique structure of the Faculty of Communication, Culture, and Society, which houses numerous researchers studying communication from various perspectives. This diversity is reflected in the Master's programmes. Additionally, I believe Switzerland presents highly appealing opportunities for researchers due to the abundant research funding available, both for fundamental and more applied research. Lastly, in Lugano, I have observed a strong interest from the local community to attend events and conferences, which I find very encouraging and rewarding." 

What are the upcoming challenges for this discipline from a scientific research perspective? 

"What I see as vital for the future is the collaboration between technology and people. We need to better understand the strengths of technologies and the strengths of people to find better ways to collaborate. Although it is now possible for AI systems to use fluent language in conversation with humans, there are certain cognitive abilities they do not master, especially those that require complex reasoning. I am collaborating with cognitive scientists from the University of Amsterdam in my SNSF project, C-LING, which investigates the ability of machines to do tasks that involve creative thinking, such as analogical reasoning and creative storytelling.

Human-AI teaming is a central theme as well in an Innosuisse project I am leading in collaboration with a Ticinese company, ARCA24, and the EHL Hospitality Business School, which specialises in recruitment. This project aims to enhance the existing system used to match suitable CVs with specific job positions. We have developed a more effective strategy utilising machine learning that operates in multiple languages, including Italian, French, Spanish, Portuguese, German, and, of course, English. The solution relates to both hard skills, which will remain present in any case, and soft skills, which are equally important in today's working world. Also, it remains important that HR professionals work together with the applicant tracking system to avoid unwanted biases being introduced in hiring processes.  

Another challenge I have been working on for many years is creating NLP tools that serve people in their mother tongue, incorporating their own cultural values. It is essential that data that is used to train NLP systems is collected from various languages, cultures, and nations around the world, ensuring that everyone is represented equally. For example, even though ChatGPT can communicate in many languages, much of the data it uses primarily originates from the US. We may be under the impression that we are conversing with someone from our culture simply because the conversational agent speaks our language. However, this is misleading since the data comes from different sources and may present the cultural values of largest data provider. To ensure the survival of a language, it is essential to have computational tools available in that language. Without these tools, the language is at risk of disappearing as it becomes less practical for everyday use. I am currently working in the NCCR Evolving Language that involves several universities across Switzerland. This project aims to study how languages evolve over time. My team focuses on how new terms are introduced in a language and how they spread in communities, for example, the term 'sharing economy'. We will compare Western societies and hunter-gatherer societies in secluded areas that are very different in nature in multiple aspects."