The AI•ON process & FAQ


Team formation

Researchers and students interested in tackling a specific problem should sign up to the mailing list (Google Group) associated with the problem or get in touch through. The link can be found next to the problem description. We also have a AI•ON gitter chat. Anyone is free to join. Teams are meant to be self-organized and self-supervised, with only occasional lightweight mentorship from senior researchers.



Problem solving

The entire research process takes place in the open, with communications happening via either a Google group or a Gitter channel. Code is exchanged via public git repositories on GitHub or Gitlab. The repositories are created and managed by team members.




The research process takes place in the open, which is already a form of publication and result-sharing. However some results may be organized into research papers to be submitted to Arxiv and research conferences. Authors lists and credit attribution is expected to be handled by team members, with conflicts escalated to third-party arbiters. Every person who made a contribution, internal or external to the team, should be getting appropriate and commensurate recognition for the value they provided. In particular, senior researchers who submitted the problem that is the topic of the publication and who provided mentorship throughout the research process should have a reasonable expectation of co-authorship. AI•ON itself should also be acknowledged in some form in all publications (e.g. under the "acknowledgments" section).



Conflict resolution authority

Openness is our core value, and an organization can only be open if power within it is transparently distributed to many individuals. However, having access to a third-party authority is still necessary to quickly resolve any conflict that may arise. AI•ON maintains a list of arbiters that may be contacted in case of issue. The list is as follow (note that it will grow as AI•ON matures):

  • Francois Chollet -
  • Gabriel Pereyra -
  • Hugo Larochelle -



On the fear of "scooping"

The fact that the research process of every AI•ON project gets publicly documented, from ideation to open-source code deliverables, may raise concerns about having ideas or code taken from AI•ON by unscrupulous parties and published as their own. This is both inevitable and utterly unimportant. Inevitable because such events always occur no matter where results come from and how they get published. Unimportant because it has ultimately no long-term consequences. The fact that the research process of AI•ON is fully public and continuously timestamped by neutral third-parties (Google, GitHub, etc.) makes it possible to easily prove where specific results originated.