PROFILE · How should automated profiling be regulated?
„Хоризонт 2020“ — Действия „Мария Склодовска-Кюри“
- Период
- 2018-01-01 → 2019-12-31
- Финансиране от ЕС
- 172 800 €
- Участници
- 1
- Схема
- MSCA-IF
Линиите свързват координатора с партньорите.
Накратко на български
Автоматизираното профилиране анализира данни, за да определя различни цени или условия за услуги, например при онлайн пазаруване и застраховки. Анализът помага да се разбере дали законите трябва да се променят, за да се предотвратят дискриминация и нарушения на човешките права.
Кратко обяснение, генерирано от езиков модел по текста на CORDIS. Оригиналът е по-долу.
Резултати накратко
How should automated profiling be regulated?
Profiling is a type of algorithmic decision-making that involves automated processing of personal or other data to develop profiles that can be used to make decisions about people. Some companies use thousands of data points to take automated and opaque decisions: online shops can sell the same good to different consumers for different prices. Lenders can set interest rates for individual consumers, or refuse to lend to them. Insurers can adjust premiums to individual consumers, or deny them insurance. Profiling advances important goals, such as efficiency and economic growth. But profiling may threaten values the law aims to protect, and goals it aims to achieve. For instance, profiling can lead to unfair or even illegal discrimination. This project examines profile-based price and service differentiation. Current law regarding profiling is unclear and may fail to protect important values. The overarching research question for the project is: to protect human rights, while considering the particularities of different sectors, should the law be amended because of automated profiling, and if so: how?
Текст от CORDIS, на английски · Данни: CORDIS, © Европейски съюз
Цел на проекта
Profiling may threaten values that the law aims to protect, and undermine goals that the law aims to achieve. Profiling involves automated processing of personal or other data to develop profiles that can be used to make decisions about people. Profiling can be used in different contexts. For instance, (i) with retail price discrimination, online shops charge different consumers different prices for the same or similar products. (ii) Lenders use profiling to estimate a consumer’s creditworthiness. Lenders can adapt interest rates to certain consumers, or refuse to lend to them. (iii) Predictive policing refers to the use of profiling technology to predict criminal behaviour.However, profiling has drawbacks. For instance, profiling can discriminate unintentionally, when an algorithm learns from data reflecting biased human decisions. Additionally, profiling is opaque: people may not know why they are treated differently. Making profiling transparent is difficult, among other reasons because of the complexity and the possibly ever-changing nature of algorithms. The project’s overarching research question is: considering the rationales for the rules in different sectors, is additional regulation needed, and if so: how should profiling be regulated? The project aims to develop guidelines for regulating profiling.I examine profiling in three sectors: retail price discrimination, consumer credit, and predictive policing. For each case study, I analyse current rules that apply to profiling. Next, I analyse these rules’ rationales, which are partly different for each sector. A rule may, for example, aim to protect a human right, or express a legal principle, such as equality, contractual freedom, or the right to a fair trial. Rules may also have economic rationales, which are different for each sector. Drawing from the three case studies, I develop guidelines to regulate profiling. Policymakers, NGOs, and other stakeholders expressed great interest in the results.
Оригинален текст от CORDIS (на английски).
Участници
- VRIJE UNIVERSITEIT BRUSSEL · Bruxelles / BrusselКоординаторБелгия
Връзки
Данни: CORDIS, © Европейски съюз