Big Data | AI & Law

• Prof dr. Haroon Sheikh

Description

This morning session of the first day will give a general introduction to societal issues relating to AI. We will look at the history of AI and see how its definition evolved in relation to technical and social developments. We will also look at utopian and dystopian images of AI in public debate and popular culture. Finally, the module will deal with patterns in the AI strategies of different countries.

Learning goals for this session are:

1. Understand different currents in AI and their evolution
2. Analyse frames of human interaction with AI
3. Differentiate common patterns as well unique elements in national AI strategies

• Prof. dr. Eleni Kosta

Description

Most companies already deploy machine learning algorithms, which are algorithms that can learn from and make predictions on data, but cannot explain their outcomes (they are a ‘black box’).  Although the European General Data Protection Regulation (GDPR) does not provide for compliance requirements specific to applying machine learning algorithms, the combined requirements of the GDPR entail that machine learning algorithms need to be designed, developed and applied in a transparent, predictable and verifiable manner (‘Algorithmic Accountability’). This is also in line with recent guidance on deployment of AI issued by the European Data Protection Board. This afternoon session will analyse in depth the provisions of the GDPR that are relevant for AI, it will discuss the draft AI Regulation and will explain what Algorithmic Accountability means in practice, including:

• Analyse the relevant data protection principles in play, such as fair, lawful and transparent processing & accountability;
• Identify the special challenges that arise with regard to these principles in the context of AI big data analytics;
• Discuss additional challenges that arise when AI involves automated decision-making based on profiling;

• Prof. dr. Lokke Moerel &  mr. Marijn Storm

Description

The morning and afternoon session of the second day will build upon the topics and issues dealt with and explained in the afternoon session of the first day. Prof. Lokke Moerel and Mr. Marijn Storm will discuss in detail the 10 steps to implement to achieve algorithmic accountability in accordance with the GDPR and the draft AI Regulation.

Participants will get insights on:

• The latest on how to address unlawful bias in algorithms
• How to deploy an algorithm to prevent existing unlawful discrimination
• Cursing in the privacy church: why we need sensitive data categories to address bias
• What do the EDPB and EU Data Protection Authorities expect and what can actually be achieved?
• How to train AI if you need to make distinctions between minority groups, like in healthcare
• When does disparate impact turn into disparate treatment?
• Alternatives to facilitate explainability to individuals if the AI is a black box
• How to document the development process and comply with the DPIA requirements under GDPR
• If we follow GDPR, are we OK under the draft AI Regulation?
• If we follow GDPR, are we also OK elsewhere in the world?
• If we follow GDPR, do we have to do more about ethics?

• mr. Frank van Praat, MA.MSc, RE

Description

Supervisory authorities around the globe, typically consider the so-called three lines of defense model as best practice for risk management and internal control. This model is not fit-for-purpose when it comes to digital innovation. Because new technologies are not fully regulated yet, it is difficult to perform a clear-cut compliance check. AI opens up a whole new range of design issues and associated ethical dilemma’s.  Years of controls by the compliance function have undermined the self-learning capacity of the business to make contextual assessments and factor in ethical considerations. This leaves the compliance department with no other option but to reject the innovation. Responsible innovation is only possible if the relevant compliance experts are part of the innovation team and if teams take joint responsibility for compliance.

• How to adapt the 3-lines-of-defense model to ensure responsible innovation
• How to train your data scientists and innovation teams on ethical dilemmas
• How to implement quality assurance and business controls to ensure ‘responsible AI’

• Ruben Roex & Hans Graux

Description

In our increasingly information-based and algorithmic society it is not surprising that the attention for Responsible AI is rapidly growing. Responsible AI is about organizations ensuring that their use of AI fulfills a number of criteria. First, that it’s ethically sound and complies with regulations in all respects; second, that it’s underpinned by a robust foundation of end-to-end governance; and third, that it’s supported by strong performance pillars addressing bias and fairness, interpretability and explainability, and robustness and security.

While theories in this domain still are in an early stage of development, practice is already placing designers, developers, users, and subjects of these AI systems before important risks, dilemmas and choices.

In this session we will dive into the practice of Responsible AI. Through an interactive case study, we will look at what it takes to develop an AI algorithm and what the key risks and considerations are in doing so. Furthermore, we will explore current approaches and instruments for AI Accountability, using examples from practice.

• Prof. dr. Linnet Taylor  

Description

The morning session of the fourth day will deal with the relevance and use of ethics in relation to big data analytics. Topics will include the relevance of ethics for privacy, accountability, transparency and fairness. We will also consider the new problems on the individual, group and societal level that are arising through data analytics, algorithms and AI, and will discuss how ethical frameworks intersect with legal ones in guiding the real-world activities of data scientists. Goals for this module are for participants to:

1. Understand the different perspectives on ethics in relation to data science, and be able  to apply these perspectives to evaluate practice;
2. Differentiate between the risks of data analytics that are addressed by current data protection frameworks and those that are not;
3. Determine which conceptual framework to apply to a given problem, and how to relate ethical principles and practice broader problems in relation to emerging data analytic techniques.

• Prof. dr. Lokke Moerel & Prof. dr. Linnet Taylor  

Description

The day will end with a discussion of ethical dilemma’s encountered in practice when developing and implementing AI tooling

• In retail (loyalty programs)
• Distribution (optimization)
• HR recruiting tools
• Fraud prevention tooling
• Financial services (inclusive financing, customer duty of care)

Course fee

The course fee is due 14 days before the start of the course. No VAT will be charged for this course.

Changes

Tilburg University reserves the right to change parts of the course in the event of unforeseen circumstances or recent developments. You will be informed of any changes as soon as possible.

Cancellation

Cancellation by a course participant needs to take place in writing. If you cancel in writing not later than four weeks before the first course day, the course fee will be refunded. Tilburg University reserves the right to cancel the course if an insufficient number of participants have registered.

Evaluation

After every course day, the course participants will be given an evaluation form. It is very important for the further development of the post-academic courses to receive the course participants’ feedback through these evaluation forms. The information obtained will be used as much as possible in organizing and designing future courses.

Complaints

The course participant can report any complaints to Tilburg University in writing. The complaint needs to be described in detail. Complaints do not suspend the obligation to pay the course fee. If the complaint is upheld by Tilburg University, the course participant will receive a reduction of the course fee.