Hello, world!
Originating as an open-source company, Hugging Face was founded on some key ethical values in tech: collaboration, responsibility, and transparency. To code in an open environment means having your code – and the alternatives inside – viewable to the world, associated together with your account and available for others to critique and add to. Because the research community began using the Hugging Face Hub to host models and data, the community directly integrated reproducibility as one other fundamental value of the corporate. And because the variety of datasets and models on Hugging Face grew, those working at Hugging Face implemented documentation requirements and free instructive courses, meeting the newly emerging values defined by the research community with complementary values around auditability and understanding the mathematics, code, processes and those who result in current technology.
Methods to operationalize ethics in AI is an open research area. Although theory and scholarship on applied ethics and artificial intelligence have existed for many years, applied and tested practices for ethics inside AI development have only begun to emerge throughout the past 10 years. That is partially a response to machine learning models – the constructing blocks of AI systems – outgrowing the benchmarks used to measure their progress, resulting in wide-spread adoption of machine learning systems in a variety of practical applications that affect on a regular basis life. For those of us excited about advancing ethics-informed AI, joining a machine learning company founded partially on ethical principles, just because it begins to grow, and just as people internationally are starting to grapple with ethical AI issues, is a chance to fundamentally shape what the AI of the long run looks like. It’s a brand new type of modern-day AI experiment: What does a technology company with ethics in mind from the beginning appear like? Focusing an ethics lens on machine learning, what does it mean to democratize good ML?
To this end, we share a few of our recent pondering and work in the brand new Hugging Face Ethics and Society newsletter, to be published every season, on the equinox and solstice. Here it’s! It’s put together by us, the “Ethics and Society regulars”, an open group of individuals across the corporate who come together as equals to work through the broader context of machine learning in society and the role that Hugging Face plays. We consider it to be critical that we’re not a dedicated team: so as for a corporation to make value-informed decisions throughout its work and processes, there must be a shared responsibility and commitment from all parties involved to acknowledge and learn concerning the ethical stakes of our work.
We’re repeatedly researching practices and studies on the meaning of a “good” ML, trying to supply some criteria that would define it. Being an ongoing process, we embark on this by looking forward to different possible futures of AI, creating what we will in the current day to get us to a degree that harmonizes different values held by us as individuals in addition to the broader ML community. We ground this approach within the founding principles of Hugging Face:
-
We seek to collaborate with the open-source community. This includes providing modernized tools for documentation and evaluation, alongside community discussion, Discord, and individual support for contributors aiming to share their work in a way that’s informed by different values.
-
We seek to be transparent about our pondering and processes as we develop them. This includes sharing writing on specific project values in the beginning of a project and our pondering on AI policy. We also gain from the community feedback on this work, as a resource for us to learn more about what to do.
-
We ground the creation of those tools and artifacts in responsibility for the impacts of what we do now and in the long run. Prioritizing this has led to project designs that make machine learning systems more auditable and comprehensible – including for individuals with expertise outside of ML – corresponding to the education project and our experimental tools for ML data evaluation that do not require coding.
Constructing from these basics, we’re taking an approach to operationalizing values that center the context-specific nature of our projects and the foreseeable effects they could have. As such, we provide no global list of values or principles here; as a substitute, we proceed to share project-specific pondering, corresponding to this article, and can share more as we understand more. Since we consider that community discussion is essential to identifying different values at play and who’s impacted, we have now recently opened up the chance for anyone who can hook up with the Hugging Face Hub online to supply direct feedback on models, data, and Spaces. Alongside tools for open discussion, we have now created a Code of Conduct and content guidelines to assist guide discussions along dimensions we consider to be essential for an inclusive community space. We’ve got developed a Private Hub for secure ML development, a library for evaluation to make it easier for developers to guage their models rigorously, code for analyzing data for skews and biases, and tools for tracking carbon emissions when training a model. We’re also developing recent open and responsible AI licensing, a contemporary type of licensing that directly addresses the harms that AI systems can create. And this week, we made it possible to “flag” model and Spaces repositories with the intention to report on ethical and legal issues.
In the approaching months, we shall be putting together several other pieces on values, tensions, and ethics operationalization. We welcome (and need!) feedback on any and all of our work, and hope to proceed engaging with the AI community through technical and values-informed lenses.
Thanks for reading! 🤗
~ Meg, on behalf of the Ethics and Society regulars
