Exploring the Intersection of Openness and AI
The rapid advancements in AI challenge the concept of openness on the internet, as companies use publicly available data to their advantage, frequently disregarding the concerns and welfare of other parties, such as artists and content creators, and the impacts of the tools they make available for use.
There is a growing realization that the existing definitions and approaches to open (as in open data or open knowledge) and open source might not be well equipped to address some of the challenges related to AI systems.
To help us make sense of this fast-paced environment, we compiled a list of questions that define the field of “Openness and AI.” Our efforts in this space aim to look into how we can preserve openness as a value that increases transparency and trust and lowers barriers to participation while accounting for its externalities and keeping potential harms associated with developing and deploying AI systems in mind.
The document includes questions about
- input/data/dataset governance,
- the impact of open AI systems on creators/creativity,
- field definitions, transparency/traceability/legibility requirements, and
- regulatory issues/AI Act-related topics.
These questions are intended to help structure discussions about AI/ML and openness and imagine a “new open” that is resilient to abuse and unintended consequences.
We are sharing this list of questions on PubPub, as an open invitation for feedback and input from others. Our goal is to foster a collaborative dialogue around these questions to improve and broaden our shared understanding of this complex field.
Anyone interested in these topics is encouraged to contribute to the discussion by sharing their own questions, answers, and points of view. This will help us develop a more comprehensive and nuanced understanding of the issues.
