GSoC 2022 Project Idea 21.1: A Collective Cognition Model for AI Ethics

The need for automated evaluation of real-time data is important in a number of sociotechnical contexts. Our group is looking to develop an auditing system and simulation of collective cognition that will improve open-source community sustainability. This interdisciplinary approach to AI ethics will involve both the development of a homeostatic system that encourages cooperative and altruistic interactions, and using simulated data generated through an agent-based model of open-source behaviors and interactions.

In taking a cybernetic approach, the candidate will build an analytical model that incorporates features such as general feedback loops (recurrent relationships) and causal loops (reciprocal causality). This might be in the form of a traditional boxes and arrows (input-output) model, or something more exotic such as Reinforcement Learning. Applicants might take inspiration from Mick Ashby’s ethical regulator (Ethical regulator - Wikipedia).

The broader goal is to build a model of cultural evolution that will encourage desired behaviors. The first part of this project will involve building a computational system to model the resources, activities, and interrelationships of an open-source community. The second part of the project will involve simulating this community using an agent-based model, which will provide the candidate with output data necessary to train and benchmark the cybernetic model.

What can I do before GSoC?
You can join the Orthogonal Lab Slack and Github, as well as attend our Saturday Morning NeuroSim meetings. You will work with our Ethics, Society, and Technology group, and interactions with your colleagues is key. You will also want to become familiar with a scientific programming approach (such as Python or Julia) to construct your cybernetic model, as well as the NetLogo platform for building agent-based models.

Requirements
Expertise or the ability to learn Python, Julia, or Kotlin (for the cybernetic model) and Scala and Java (for the agent-based model). The ability to extract model representations from complex systems is helpful. Knowledge of open-source development practices and an interest in interdisciplinary research are a must.

Planned Effort
350 hours. Mentors: Bradly Alicea (bradly.alicea@outlook.com), Jesse Parent (jtparent2018@gmail.com).

3 Likes

Hi,

I read the project description and am extremely interested to participate in GSoC with this.
I wanted to know if you would accept people who are not currently students (I graduated in 2020 and have left my job so I have been working independently since a few months).

I have had various jobs and research internships where I worked on using ML to solve different types of problem statements, so I am very interested in interdisciplinary research. Also, although I have developed a curiosity for computational neuroscience, I don’t have any past experience in neuroscience.

Let me know what steps I need to start with if I am someone you consider eligible. Cheers!

Thanks for contacting me. The GSoC program does accept non-student for the first time this year! If you are interested in more information about this project, please join our lab Slack: Orthogonal Lab Please check your Inbox and Spam folder for the invite, then join the #ethics-society-tech channel. I will be discussing this project in the coming week with potential applicants. Look forward to seeing you there!

1 Like

I’ll be there, thank you! Looking forward to it!