After this course, the student is able to: 1. Understand the ethical challenges involved in working with language technology 2. Identify the ethical challenges posed by NLP research and how to address them in practice and in theory (in models and in papers) both as developer and as user/reader 3. Report on ethical issues (such as data statements, bias statements, etc) 4. Be aware of the state-of-the-art discussion on ethics in the NLP community 5. Reflect on sociodemographic aspects of language and on the tradeoff between successful research and legitimate ‘behaviour’. 6. Reflect on what is responsible behaviour and the dangers of potential third-party misuse 7. Understand the importance of model interpretability, and know which techniques are available for achieving it, for detecting bias in models, and for debiasing models (2.1, 2.3)
Prerequisiti
As prerequisites, we expect some good knowledge of NLP and some general knowledge of state-of-the-art techniques, but there will be no programming/development involved.
Metodi didattici
The course will be articulated in a number of lectures and several laboratories where students will work on practical assignments, often in small groups.
Verifica Apprendimento
3-4 Assignments (usually in groups)
Reading/listening/watching materials individually and in small groups with open discussions/debates and presentations by the students. Materials will be a mix of theoretical, technical, popscience. Coding understanding will be an advantage.
Final essay/exam
Testi
Some reading recommendations will be provided in advance, while others will be suggested along the way, also in line with what will be recent developments at the time of teaching.
Full reference list (only a selection will be chosen, of course): https://drive.google.com/drive/folders/1lN8S0QOJSkBWOom0fq8ZaYH6y_xtr_V3?usp=share_link
Contenuti
The spread and democratisation of language technology has made it possible to use Natural Language Processing in a variety of applications. Language-based tools are indeed not only developed within academia, but also used by very many companies, large and small, far beyond just research purposes. Working with Natural Language Processing, now more than before, involves ethical reflections in many directions. These concern: (i) the choices we make when developing methods, models, and data for language processing (for example: annotation categories, features, etc); (ii) the biases that are intrinsic to human-produced data and thus to data-derived models, including explainability issues; (iii) the consequences of our work, in terms of personal responsibility and third-party (mis)use.
