Old Proposed Location: Guide for Ethical Research/Ethics-Informed Licensing New Proposed Location: Guide to Reproducible Research/Licensing Background Ethics-informed licenses have evolved from traditional open source licenses in recognition of the use of free and open source software in harmful or unethical applications. While often times these licenses build off of open licenses as a foundation, they often include restrictions based on an ethical value such as a general "do no harm" or more targeted value such as limiting use for law enforcement applications. Limitations Adoption has been limited for these licenses so far for various reasons. Some open source developers do not agree with adding restrictions on use, as that would compromise the principles of "openness". Additionally, enforcement of ethical conditions can be challenging from a legal standpoint. Ethical Source

Useful links BigScience Data Governance Framework The Stack Am I In the Stack? ODC Repo TTW Personas Issue What does the term “Data Custodian” mean to you? Do you consider yourself one? Someone who thinks about the data holistically and not just steps on the way to a research project but preparing the insfrastructure to better prepare for all projects

Yann Sweeney - DCE RAM [TOC] Useful links RAM repo RAM repo: onboarding folder Overview

Zoom: https://turing-uk.zoom.us/j/94968554438?pwd=TzJPczIvRnJHVjlFNUZ6TVIrTnlQUT09&from=addon [TOC] Useful links dobson-eq Github 28 March & 14 Oct meeting notes Agenda

https://twitter.com/turingway/status/1513541747095314441 ⚠️🚨NEW CHAPTER ALERT! 🚨⚠️ A new sub-chapter on Machine Learning Model Licenses is now live, co-written by @jen_gineered and @Carlos_MFerr! 👉 https://the-turing-way.netlify.app/reproducible-research/licensing/licensing-ml.html 👉 https://the-turing-way.netlify.app/reproducible-research/licensing/licensing-ml-case-studies.html These sub-chapters sit in our Guide for Reproducibility, covering: 👉 Definitions of ML model licenses, and how they differ from data or source code licenses 👉 Examples of existing ML models and licenses

October 4, 2022 Proposed Agenda Intro to REG (CRS/AH) Research background without individual research agenda Help make the research output open, reproducible Not just here to wrap up code, interested in contributing to the research side Way we Work: work well in sprints, Github (e.g. project board)

Emilie is a climate scientist (PhD in weather/climate modelling) Focus now on applied research (translating MO work into user applications) Worked in air pollution, wind energy MO group not sector-specific (e.g. transport, wild fires) Joined Joint Centre last year (Exeter + Met Office) James Salter RB: using synthetic populations, projecting to future, with climate/heat data to help local authorities plan adaptation stratgies

Location: Guide for Reproducible Research/Licensing/Machine Learning Licenses Like a software license, a Machine Learning (ML) model license governs the use, redistribution of the model and/or algorithm, and distribution any derivatives of it. However, there are other components to an AI system, such as data, source code, or applications, which may have their own separate licenses. ML model licenses may restrict the use of the model for specific scenarios for which, due to ethics-informed concerns, or technical limitations of the model informed by its model card, the licensor is not comfortable that the model is used. or allocate liability associated using the model, its component parts, and its outputs. While many ML models may utilise software licensing models (e.g. MIT, Apache 2.0), there are a number of ML model-specific licenses that may be developed for a specific model (e.g. OPT-175B license, BigScience BLOOM RAIL v1.0 License), company (e.g. Microsoft Data Use Agreement for Open AI Model Development), or generally applicable to a wide range of models (e.g. BigScience OpenRAIL-M (Responsible AI License)). In summary, the growing list of ML licenses reflects the understanding that the ML model and other artefacts are distinct from the code, and thus in need of new licensing options. Reproduction and propagation of ML models Many similar or related versions of a model may exist, whether it is the evolution of a model family (e.g. GPT-2 and GPT-3) or implementations of the model in different languages (e.g. DALLE vs. DALLE-pytorch). Each version may have is own license, though some model developers are now requiring all downstream models to at least have the same base license as the original. In the most extreme cases, a model developer may choose to grant the license exclusively to a single entity or open every aspect of the model to the public domain.

Location: Guide for Reproducible Research/Open Research/Open AI What is open AI? Open source has been an important attribute of AI communities in academia and industry. In the early 2000s, open source machine learning software libraries like scikit-learn and open datasets like ImageNet helped generate interest and set standards for the community. While these early projects were driven to open to build a shared and optimised resource for a community of collaborators, new models and motivations for open AI have developed. Benefits of open AI Collaboration through shared datasets & eval benchmarks Encouraging competition for innovation Growing user base and developer team

Proposed Work Guide for Ethical Research/Ethics-Informed Licensing Eirini & Jen Guide for Reproducible Research/Licensing/Machine Learning Licenses Jen & Carlos

March 13, 2022 01:00-02:00PM EST Zoom Link: https://us02web.zoom.us/j/86830805961?pwd=RytxNTBwWW9yOHBobUp2U0xjN3h5Zz09 [TOC] Agenda Time