Recently I was one of 100+ people who registered for the webinar “In Conversation with Thomas Padilla”. A video from the webinar is now available from University of Adelaide Library YouTube channel.
Thomas Padilla is the Interim Head, Knowledge Production at the University of Nevada Las Vegas. He is the author of the OCLC Research report Responsible Operations: Data Science, Machine Learning, and AI in Libraries and the lead for significant Mellon Foundation supported work in the US on “collections as data”.
Joining Thomas as organisers and facilitators of the session were:
- Ingrid Mason, independent consultant on research infrastructure and heritage data collections
- Alexis Tindall, Manager Digital Innovation, University of Adelaide Library, and
- Adam Moriarty, Head of Collection Information and Access, Auckland War Memorial Museum
- Gene Melzack, Data Curator, Student and Scholarly Services, University of Melbourne.
- Project Aida: applying image analysis and machine learning in digital libraries of historic materials
- On the Books: Jim Crow and Algorithms of Resistance: a text mining project investigating racially-based legislation in North Carolina (1865-1968).
Theme 1: Library values and ethics
We need to be aware that data is not just data; data is about lives and can reflect histories of oppression. Increasingly there are concerns about responsible use, particularly of indigenous collections, and these issues are amplified when machine learning is applied. Common frameworks such as FAIR and organisations such as the Research Data Alliance are not adequately addressing these concerns.According to the Global Indigenous Data Alliance:
The current movement toward open data and open science does not fully engage with Indigenous Peoples rights and interests. Existing principles within the open data movement (e.g. FAIR: findable, accessible, interoperable, reusable) primarily focus on characteristics of data that will facilitate increased data sharing among entities while ignoring power differentials and historical contexts. The emphasis on greater data sharing alone creates a tension for Indigenous Peoples who are also asserting greater control over the application and use of Indigenous data and Indigenous Knowledge for collective benefit.Commercial AI tools also need to be approached with caution, and Thomas warned against being “lured by scale” in ways that compromise our values. An example of this could be re-purposing a tool for cultural heritage use that was originally developed by governments for facial recognition of protestors. Adam also highlighted issues around the hidden labour involved in services like Amazon's Mechanical Turk and noted that just because something seems a cheap and easy option does not meant that it is something that cultural heritage organisations should jump on board with.
Further reading:
- Responsible Operations: Data Science, Machine Learning, and AI in Libraries
- Collections as Data: Part to Whole
- Rumman Chowdury - The Data Scientist Putting Ethics into AI
- CARE Principles for Indigenous Data Governance
Theme 2: Building multidisciplinary teams
Thomas referred to the work being done by Nancy McGovern on radical collaboration in research libraries. According to McGovern:The concept of radical collaboration means coming together across disparate, but engaged, domains in ways that are often unfamiliar or possibly uncomfortable to member organizations and individuals in order to identify and solve problems together, to achieve more together than we could separately.Thomas suggested that projects need to be intentional about providing opportunities to draw on the expertise of everyone on a project team. He observed that tech folk often "consult" non-tech contributors at the start of a project but then go off and do their own thing. Participants instead need to design the project to deliberately bring conversations between people to the fore, and to build trust and mutual respect.
Thomas also talked about being "separated by a common language". This occurs when different professional groups use the same words to mean different things (e.g. a humanist will have a different idea of scale from a machine learning specialist). How do we bridge those gaps in language and culture between different professional groups? We need to be explicit about assumptions and agree on terminology as part of setting projects up.
In terms of roles and competencies, Thomas noted that organisations need someone who can be a translator between different groups. But it can't just be that translational person's responsibility, as this is not sustainable. Organisations need to shift more broadly, and this requires leadership and managerial efforts to build a culture of collaboration and innovation. There are also questions around how to retain people who are confident and competent in this emerging area; what can we do to ensure people want to come and work with GLAM institutions rather than applying these skills elsewhere?
There was some discussion about building career paths and skills within institutions, and the pros/cons of outsourcing vs building internal capabilities vs a combination of these. In terms of building internal capabilities, Thomas mentioned at least three strategies:
- The Carpentries. There is now a strong body of evidence that those pedagogies work and they are also reasonably affordable.
- Collections as Data: 50 Things You Can Do. A list compiled by the Always Already Computational project, of 50 things staff in cultural organisations can do to “open eyes, stimulate conversation, encourage stepping back, generate ideas, and surface new possibilities” in relation to collections as data.
- Workplace learning. Thomas gave the example of a staff professional development approach at Michigan State University Library. In consultation with supervisors, any staff member could devote 25% of their time could to shadowing or cross-team projects. This was re-evaluated as part of the annual performance review process but could be ongoing. This led to cross-fertilisation, up-skilling, and was a great way for people new to the profession to try different things.
Further reading:
Radical Collaboration and Research Data Management [special issue]. Research Library Issues, no. 296 (2018).
About AI4LAM
The webinar was the first public event held to gauge the level of interest in establishing an ANZ chapter of AI for Libraries, Archives, and Museums (AI4LAM), “an international, participatory community focused on advancing the use of artificial intelligence in, for and by libraries, archives and museums.” Further information about AI4LAM is available via their website.The organisers of the local webinar are a small group of professionals interested in the role of computational and curatorial techniques that advance artificial intelligence in cultural heritage practice and allied areas in research (e.g. digital humanities).
You can register interest in participating in future events and efforts to establish the local chapter via an online form.