When Stochastic Parrots Learn to Swim: The Regulation of General Purpose Artificial Intelligence in the EU
(2023) HARN63 20231Department of Business Law
- Abstract (Swedish)
- General Purpose AI systems like ChatGPT pose unique dangers to society owing
to the rapid evolution in their capabilities, and their ability to be utilized for a large variety of purposes, including those unintended by the developers of these AI systems. This thesis examines those dangers, and critically analyses the EU’s draft AI Act and its amendments with a view to ascertain if the draft law adequately tackles those dangers. It finds that the draft AI Act is ill-equipped to effectively regulate general purpose AI systems, and suffers from issues of rigidity, onerous regulatory obligations, and a lack of consideration of the unique characteristics of general purpose AI systems. The thesis explores and critiques the solutions offered by... (More) - General Purpose AI systems like ChatGPT pose unique dangers to society owing
to the rapid evolution in their capabilities, and their ability to be utilized for a large variety of purposes, including those unintended by the developers of these AI systems. This thesis examines those dangers, and critically analyses the EU’s draft AI Act and its amendments with a view to ascertain if the draft law adequately tackles those dangers. It finds that the draft AI Act is ill-equipped to effectively regulate general purpose AI systems, and suffers from issues of rigidity, onerous regulatory obligations, and a lack of consideration of the unique characteristics of general purpose AI systems. The thesis explores and critiques the solutions offered by scholars to remedy these issues, and finally proposes an alternative regulatory approach focused on a goal-based regulation that would empower the European Commission to frame and amend rules governing general purpose AI systems, as well as the AI space at large. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9122921
- author
- Srivastava, Vartul LU
- supervisor
- organization
- course
- HARN63 20231
- year
- 2023
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- Artificial Intelligence, AI, GPAI, AI Act, General Purpose AI, ChatGPT, EU AI Act, AI Regulation, European Union Artificial Intelligence Act
- language
- English
- id
- 9122921
- date added to LUP
- 2023-06-19 15:30:23
- date last changed
- 2023-06-19 15:30:23
@misc{9122921, abstract = {{General Purpose AI systems like ChatGPT pose unique dangers to society owing to the rapid evolution in their capabilities, and their ability to be utilized for a large variety of purposes, including those unintended by the developers of these AI systems. This thesis examines those dangers, and critically analyses the EU’s draft AI Act and its amendments with a view to ascertain if the draft law adequately tackles those dangers. It finds that the draft AI Act is ill-equipped to effectively regulate general purpose AI systems, and suffers from issues of rigidity, onerous regulatory obligations, and a lack of consideration of the unique characteristics of general purpose AI systems. The thesis explores and critiques the solutions offered by scholars to remedy these issues, and finally proposes an alternative regulatory approach focused on a goal-based regulation that would empower the European Commission to frame and amend rules governing general purpose AI systems, as well as the AI space at large.}}, author = {{Srivastava, Vartul}}, language = {{eng}}, note = {{Student Paper}}, title = {{When Stochastic Parrots Learn to Swim: The Regulation of General Purpose Artificial Intelligence in the EU}}, year = {{2023}}, }