Ongoing Research

Worldwide AI Ethics is a systematic literature review done by AIRES researchers. Building on the work done by other meta-analysts, this study presents an analysis of 200 documents related to AI ethics and governance, presenting a collection of typologies used to classify our sample, all condensed into an interactive, freely accessible online tool.

Extend the Worldwide AI Ethics sample. We intend to keep on making AIRES the number one meta-analysis of the field. For this, we need to keep updating our dashboard with new documents. The research involves searching and analyzing documents (AI Ethics Guidelines, Bills, Legislation, Governance Frameworks, Codes of Conduct) related to AI Governance. Documents and their respective information are submitted via google forms to a central spreadsheet for further analysis and post-processing. We would like to make new publications as soon as we have a bigger and more representative sample size (e.g., 300 documents). We would like to have the help of people who understand different languages, so we can analyze documents from different places and cultures.

People wanting to learn about AI Ethics, AI governance, and AI regulation. Students from the fields of Law, Philosophy, and Social Sciences were the main contributors to the first iteration of this research. Since this is a review, it's a great way for anyone that wants to learn more about the field to "get started."

We are currently building a library of published machine learning models. This library contains information about these models, their capabilities, and the possible risks (together with ethical concerns) regarding their development/utilization. The aim is to map the biggest problems and threats that contemporary AI poses to our society.

This research involves searching for AI systems currently being used (especially the most proficient). You will have to read and learn about their workings, how they were developed, and how they operate, so you can create a preliminary risk/ethical evaluation. Information is usually found in the form of publications, code repositories, model cards, and similar sources.

People interested in learning about state-of-the-art in Machine Learning, Robotics, and AI system development. Students from any field can learn, but people with more technical backgrounds (e.g., Computer Science) will probably have an easier time understanding the papers and technical descriptions. Since this is a review, it's a great way for anyone that wants to learn more about the field to "get started."

We are currently building a web application to automate an experiment in moral philosophy. This experiment aims to find out how ethical guidelines influence software developers' decision making. We also are seeking to investigate different forms of how we can "teach ethics" in STEM fields.

This research involves applying the tool of our research. Showing it to possible respondents. Conducting controlled surveys, and similar tasks.

Experimental research is usually a little bit more tricky. We would like to encourage anyone who wants to help to join efforts with us (there are activities besides data collection you can help with), but if you want to help with the experiment design and data collection, some experience in experimental research would be good.

The Teeny-tiny Castle is a repository of educational tools for AI Ethics and Safety research. It is a python based course on how to create and use tools for addressing certain safety issues in AI (e.g., interpretability, sustainability, fairness, robustness). In it, you can also find an introductory course on ML (Machine Learning).

This research involves programming and software development in Python. It is basically the job of training, evaluating, and attacking ML models. Showing how to do this in a clear and instructive way. We should also look for new libraries and tools to use. And, in the end, convert all accumulated knowledge into a GitHub repository.

People with experience in Python, ML, Data Science, and programming in general. Cybersecurity enthusiasts are welcome!

Aira is a chatbot. Ai.ra is designed to simulate the way a human (expert) would behave during a round of questions and answers (Q&A). Aira has many iterations, from a closed-domain chatbot based on pre-defined rules to an open-domain chatbot achieved via instruction-tuning. Aira has an area of expertise that comprises topics related to AI Ethics and AI Safety research. We are currently looking to increase the reach of our system and make it an even better tool. The Aira-Instruct series was developed to help researchers explore the challenges related to the Alignment problem.

Increase the domain expertise of our Language model. Adding more languages is something we would like to implement. The activity mainly consists of improving the datasets, training, and adding new models to the Aira series.

People who master more than one language, and have an interest in working with areas related to AI Ethics and Safety. Individuals with expertise in NLP and language modelling are very welcome!

We are currently developing two types of tools: EPS (Ethical Problem Solving) and a series of AIAs (Algorithmic Impact Assessment). These tools are made to troubleshoot developers and help them create systems in a human-centric way (value-sensitive design). AIAs are intended to evaluate the potential risks of an AI application (ex-ante).

Studying existing tools, and helping other researchers to develop new and better methods to incorporate ethics inside AI system development.

People that are interested in creating tools like surveys and self-assessment questionnaires. All areas of knowledge are welcome in this endeavor, especially the Humanities and STEM fields. People with very specific backgrounds (e.g., psychology) can help to create tools aimed at assessing AI applications aimed at specific areas (e.g., mental health).