Changes and Challenges in Heritage and Open Knowledge
Download Executive Summary: Changes and Challenges in Heritage and Open Knowledge (Word file 86kb).
Introduction
Heritage institutions are increasingly rethinking their approaches to preservation, access and digital collections, following the technical and ethical challenges of AI and the ongoing work of marginalised peoples to ensure cultural practices are representative.
These complex, overlapping and emergent areas are significantly reshaping how the cultural sector thinks about effective and inclusive open knowledge and open cultural practice. The purpose of this event was to capture the current state of play across the UK cultural sector, with attendees at different professional and organisational stages exploring challenges, opportunities and questions.
Purpose
The primary aim of this event was to explore how technological innovations, particularly AI, and changes in open knowledge practices are influencing the preservation, sharing, and ethical management of cultural heritage. Key topics included the implications of AI training on heritage collections, equity and ethics in the management of Indigenous and marginalized community data, and strategies for aligning institutional practices with broader equity and access goals
Context
The roundtable was convened at a critical time when advancements in AI and the growing emphasis on open access to knowledge are pushing heritage institutions to rethink their approaches to preservation, access, and the ethical use of digital collections.
Objectives
This report aims to capture the key discussions and recommendations from the roundtable event, including reflections from the panellists and participants. Comments from the participants have been summarised and are unattributed.
The central themes discussed included the impact of AI on open licensing, ethical considerations in the management of Indigenous and marginalized community data, and strategies for aligning institutional practices with broader equity and access goals. These themes were set within the context of the ‘Paradox of Open’, the ‘CARE’ principles, and Digital Black Practice, case studies and observations from speakers and panellists.
Participants
Attendees had various levels of knowledge and experience in AI (artificial intelligence) and IP (intellectual property) and represented a diverse group of stakeholders, including:
- Heritage professionals from museums, libraries, and archives
- Academics specializing in digital humanities and data ethics
- Technologists working on AI and machine learning
- Open knowledge advocates promoting equitable access to information
Event structure
The event was chaired by Josie Fraser, Head of Digital Policy at The National Lottery Heritage Fund and was divided into two sessions one before lunch and one after. Each session was framed by a keynote presentation introducing four provocations, followed by an expert panel and roundtable discussions. The first session was on Open Licensing and Artificial Intelligence (AI) and the second session focused on Equity and Ethics in Open Licensing.
Session 1: Open Licensing and Artificial Intelligence (AI)
Keynote: Alek Tarkowski
Panellists: Alek Tarkowski, Dr Aruna Bhaugeerutty and Dr Mathilde Pavis
Provocations and roundtable discussions
Session 2: Equity and Ethics in Open Licensing
Keynotes: Dr Kirsten Thorpe, and Kelly Foster
Panellists: Dr Kirsten Thorpe; Kelly Foster; Somaya Langley and Dr Andrea Wallace
Provocations and roundtable discussions
This report reflects the structure of the event and captures the key discussions and recommendations, including reflections from the panellists and participants. Comments from the participants have been summarised and are unattributed. The content in the report reflects the tougher conversations undertaken after lunch. This also reflects that there is a lack of skill, experience and confidence in this area within the sector.
Session 1: Open Licensing and Artificial Intelligence (AI)
Keynote summary
Alek Tarkowski, Director of Strategy at the European digital rights thinktank, Open Future discussed how AI is trained on heritage digital collections and data. He noted that this has prompted important questions about licensing, ownership, and ethics. Alek discussed Open Future’s influential position paper ‘The Paradox of Open’ (2021).
The Paradox of Open refers to the challenge of balancing the benefits of open access with the risks that openness poses. While the core philosophy of open knowledge is to make resources freely available to the public, this approach can sometimes lead to the exploitation of these resources, particularly in contexts where commercial interests, like AI developers, capitalize on open data without contributing back to the cultural commons. This paradox becomes even more complex when heritage collections, especially those of marginalized or Indigenous communities, are freely accessible and used in ways that may conflict with the values or expectations of those communities.
The report suggests the concept of a more mindful approach to openness in which not everything should be open therefore achieving a better balance between open and privacy. https://paradox.openfuture.eu
First provocation: Are restrictions to sharing desirable, or necessary, to protect the commons in the context of AI?
A major push over last few decades towards digitising and publishing collections online to make them publicly available has led to a major shift for cultural heritage sector professionals – essentially becoming a content provider. It has made it increasingly important for us to grapple with aspects of rights and risk management. Within this context, many of us are already selecting what we make available online, to ensure the integrity of data put out there, compliance with copyright law etc. The advent of AI technologies reinforces the importance of implementing and extending these kinds of policies and processes. We will need to take a “risk-aware approach” rather than defaulting to unrestricted sharing.
Dr Aruna Bhaugeerutty
The devil is in the details for me, in particular the basis or reason for the restriction, and how consistent the vision on AI is with that position. If it is an issue of rights infringement such as privacy. yes. If it is an issue of problematic content that needs to be well contextualised before it is engaged with, probably yes … If it is an issue that you fear other organisations (with commercial applications in particular) will free-ride upon your investment in digital collections without sharing that benefit with you (for example via payment), then I would give more thoughts to this reasoning in light of your mission, the purpose of making collections available online, the source of that funding, the costs to you (as an organisation) of holding that line, and see if this position tracks (i.e. remains aligned with your values, your mission and evidence).
Dr Malthide Pavis
Case for restrictions
There was a broad consensus among participants that some restrictions to how materials are shared online could be introduced, and still be “in the spirit of openness”. AI is not sophisticated enough to manage its output yet, leading to remixed content, infringed copyright, out of context usage of content, drawing on only what is there and therefore creating biased outputs. However, participants doubted permissions could be managed through existing copyright laws or open licenses.
Case against restrictions
AI relies on both the quantity and quality of the data on which it is trained. By restricting access to data there is a risk that the quality of the content that AI is trained on will suffer as a result. Cultural heritage organisations need to focus on the quality of their data and ensure that it does not reinforce cultural biases, colonial perspectives and inequalities. The more and varied content that is ingested by AI, the better the outputs.
Types of restrictions
Participants explored whether imposing restrictions on the sharing of digital collections, such as only making certain types of content available online might be necessary to prevent the exploitation of cultural heritage for commercial gain. However, restricting access might be seen as censorship, for disadvantaged people, including communities wanting to explore their own heritage.
Participants also discussed that restrictions may also be necessary due to the impact on cultural heritage organisations’ digital infrastructure due to the increased traffic on websites from AI companies’ data mining activities. The example of Wikimedia Enterprises was discussed as an example of responsible sharing of data, where an API (Application Programming Interface) is provided with a service level agreement (SLA).
Whose access should be restricted?
The provocation raised the question of whose access to cultural heritage websites should be prioritised individual users or AI companies? Participants also raised wider fundamental questions such as who has the right and authority to share? Who are the beneficiaries? Can there be a single rule? Is it on a case-by-case basis? Who are the arbiters? Where does reparation fit into this?
Organisational requirements and access
Data cleaning is crucial but resource heavy. Investment is needed in the digital infrastructure of heritage organisations to support access as well as the need for investment in training so that staff, volunteers and trustees have the required digital literacy and skills. Workflows and processes that build in standardised, interoperable management of shared digital resources will save the UK cultural sector significant time and money, as well as protect resources for future generations.
Open Access, Creative Commons and AI
Several questions were raised relating to this topic. What are the mechanisms for cultural heritage organisations to implement restrictions that prevent AI companies from exploiting their data but still permit other forms of open access and re-use exemplified by Creative Commons licences? If Creative Commons licences permit the commercial use and adaptation of digital assets, then is it right for cultural heritage organisations to try and restrict the use by AI companies and those using their AI services to create new works?
Restrictions, access and ethics
Cultural heritage organisations have licences in place with copyright holders that were drafted and agreed before AI became an issue. Is it ethical for cultural heritage organisations to allow unrestricted access by AI companies to works protected by copyright when no consent has been granted by the creators? Should the creators of works have the right to make decisions to opt in or out rather than the institutions that own their works? Should the default option be to opt in therefore requiring informed consent of rights holders rather than opt out? How can AI tools be developed to ensure transparency and sources and rights holders are attributed? Is relying on licensing and copyright the solution or is the onus on governments and AI companies to develop an ethical regulatory framework in which AI companies operate? What role should the cultural heritage sector play in formulating an ethical framework?
Second provocation: There is much conversation about payments for using various resources for AI training. Should heritage institutions benefit from sharing their collections? In what way?
I think it is worth exploring further, given the funding challenges faced by the sector, especially for fundamental services that will enable open knowledge such as digitisation. AI developers need authoritative, well-structured datasets to improve their models. GLAM (galleries, libraries, archives, and museums) institutions have this in abundance – we have cultural heritage collections data comprising of millions of person-hours of knowledge production – we shouldn’t underestimate the value of this potential of licensing deals with AI companies like those made by publishers.
Dr Aruna Bhaugeerutty
Leveraging access to digital collections for training purposes in exchange for cash payment is very unlikely to happen (even if the law was in your favour). But heritage organisations may be able to leverage value in other formats (investments in-kind in their digital transformation activities, in exchange for access to datasets). … For example, have the AI company, digitise and format your digital collections for you, and ask to retain the rights, so you can publish in open access after, and they have a copy for their own purposes. Or ask for expertise, tech and facilities that advance your digital transformation. These types of exchanges are more likely to happen and be more worthwhile, than traditional licensing models.
Dr Malthide Pavis
Basis of payment
The use of licences to permit commercial use of AI raised questions on how the fee arrangements would work, and who would benefit from the licence fees: heritage organisations or creators of the works or both? This also raised issues around the relative strength heritage organisations in negotiating terms with AI companies. Other topics included: how do you quantify the value of data? Heritage organisations need to learn from other organisations that have successfully negotiated agreements on access to and use of their data. Payment models might benefit larger heritage organisations that have more data available, but how would smaller institutions benefit?
Non-financial renumeration
There was debate about whether and how heritage institutions could benefit from sharing their collections for AI training, whether through financial compensation or other means. Some argued that if institutions provide their resources for free, they should at least be acknowledged or rewarded when AI companies use their datasets. Non-financial benefits of collaboration between heritage organisations and AI companies in developing AI tools should reflect ethical issues including ensuring the attribution of sources. Other ideas also included increased funding from DCMS (Department for Culture, Media and Sport) as a reward for providing access to their data and/or governments could levy taxes on big tech companies. This raised the question of how the revenue derived from taxation would be distributed equitably?
Third provocation: What should institutional policies for sharing digital collections for AI training look like?
Current policies often tend to assume individual users, not corporate or AI users and there is a need to rethink commercial use restrictions to allow for AI training (if taking a risk-aware rather than risk-averse approach). There might also be other policy considerations such as selective sharing – select partnerships with AI actors who share our institution’s values and will help our mission put in place contracts/agreements to ensure compliance with existing laws e.g. copyright exceptions for text and data mining; transparency measures – so that it is clear about what is used, by whom, and for what purposes?
Dr Aruna Bhaugeerutty
Regardless of whether you want to allow or disallow training on your online platforms (based on your mission, values and take on open access), it is important to make sure your approach is consistent with your Terms of Use, and any other technical measures you use to communicate your policy with the public (for example, in the metadata; or robot.txt).
Dr Malthide Pavis
Policies and the Paradox of Open
Participants suggested that institutions may need to update their policies, not only to clarify how collections can be used for AI training but also to balance the desire for openness with the need for protection against commercial exploitation.
Policy content
AI policies will need to evolve as AI evolves, both in terms of what is ingested and subsequent shared via AI. Policies will also require an inherent ethical dimension to counter the ‘Paradox of Open’ and manage biases that will reflect both who has created the metadata the collecting/digitisation policies of heritage organisations. Cultural institutions should ensure they have policies in place to ensure that data is clean and meets the agreed standards before sharing with AI.
Testing policies
Heritage organisations could test their AI policies by developing their own pilots (using AI technology to make digitisation project workflows more efficient) and include built-in the human verification of AI outputs.
Fourth provocation: how can the heritage sector benefit from AI technologies? Thinking beyond the issue of open collections being used for AI training, what are public interest uses of AI?
[These include] Business operations, administration tasks (minutes etc), big data analysis: Collections – metadata creation and enrichment. Audiences – chatbots/FAQs, talking sculptures, collections discovery (via natural language querying).
Dr Aruna Bhaugeerutty
AI is best at efficiency gains for automatic labour-intensive, repetitive tasks. And there are a lot of those in preparing digital collections (like adding metadata, ALT text for example). AI can be used to connect different collections through semantic links using object detection or metadata, going beyond the classifications of the catalogue.
Dr Malthide Pavis
Collection uses of AI
While acknowledging the risks of AI misuse, there was also recognition of the vast potential for AI to support public interest uses, such as improving access to and interpretation of collections. AI technologies can enhance searchability, automate cataloguing and creating new forms of engagement with cultural materials. Other examples included machine-learning systems that can recognise text in hand-written manuscripts including transcribing and OCR (optical character recognition), as well as image analysis for collection management.
Non collection uses of AI
There is so much potential to free up staff time as well as progress with new discoveries. There is no reason why the heritage sector couldn’t be developing its own AI tools to serve its communities. AI support in areas not related to collections management could include marketing, and general museum research, outreach and supporting community engagement.
Challenges
- The Paradox of Open: how to balance the resources needed to maintain public access and openness with commercial use of cultural heritage which ‘puts nothing back’.
- Lack of understanding and regulation on how AI should interact with open collections, particularly in terms of compensation and ethical use.
- Given how current licencing compensation agreements with the creative industries, for example, across publishing and the music industry it is not clear how heritage organisations and workers would benefit from similar agreements being made in relation to AI use of collections or how public value and access to UK heritage would be supported. We know that in practice it’s difficult to get adequate compensation through this route.
- Knowledge institutions like libraries and universities have struggled to make licensing deals on behalf of the public to increase access. Currently, the law favours rightsowners, knowledge institutions are often lack the expertise and resource to make licencing deals for themselves, and aren’t able to ensure deals would be consistently applied for public benefit.
- Lack of AI legislation and regulation at national and international levels that protects and promotes public access to digital heritage, or recognises the interests of digital public spaces and resources.
- The heritage sector (like the rest of the creative industries) is currently under-served by AI tools. AI companies are investing primarily in business practice efficiencies and commercial services. Specialist services and tools are lacking – this will change as the field matures and public investment is made in AI for public good.
Opportunities
- AI technologies can dramatically improve the accessibility and searchability of cultural collections, leading to broader engagement and discovery. A key observation was “What AI can do for us, rather than just the other way around”. A key question is what AI can do to increase high-quality access to heritage and improve the heritage sector, not just what the heritage sector can do for AI.
- Developing AI-related policies that address the Paradox of Open and the need to provide the public with access to their heritage, institutions can think strategically about their digital sustainability.
- Licensing deals to train on content have already been made for access to content which could provide the basis for heritage organisations.
- Heritage institutions have an opportunity to use AI to foster public-interest applications, such as enhanced digital archives and new tools for cultural education and preservation. AI could provide opportunities for the public, volunteer and small organisations to more actively participate in the development of UK heritage collections online, for example through contributing to information about collections and their interpretation.
Session 2: Equity and Ethics in Open Licensing
Keynote Speaker: Kirsten Thorpe
Kirsten Thorpe’s keynote focused on Indigenous data sovereignty, the rights of Native nations and Indigenous Peoples to govern the collection, ownership and application of their own data, and the CARE Principles for Indigenous Data Governance (CARE). Since its publication in 2018, CARE has been increasingly used to assess whether open data and licences are applied ethically and inclusively. They reference and introduce additional equity considerations to the globally established FAIR principles (2016). The FAIR principles (Findable, Accessible, Interoperable and Reusable) are designed to ensure equitable access and availability to scientific data and have since been expanded to benefit data in its widest sense. CARE problematises and enhances the cultural sector adaptation of FAIR, by insisting the historical and community context of digital heritage assets are considered, along with how ideas of ownership are culturally situated. Kirsten drew attention to the ways in which CARE calls for heritage institutions to reimagine their practices in ways that promote Indigenous voices, ethical data management and mutual accountability.
The CARE Principles (2018) stand for Collective Benefit, Authority to Control, Responsibility, and Ethics. They were created by indigenous peoples to ensure that communities of origin have control over their own data and that their cultural heritage is protected in a way that reflects their values and interests. The acronym stands for:
- Collective Benefit: Data should be used in ways that benefit Indigenous communities, respecting their knowledge and cultural contributions.
- Authority to Control: Indigenous communities should have control over how their data is collected, used, and shared.
- Responsibility: Researchers and institutions are responsible for ensuring the proper stewardship of Indigenous data.
- Ethics: The use of data must be governed by ethical considerations, prioritizing respect for Indigenous cultures and their rights to sovereignty.
First provocation: How are UK heritage organisations engaging with the CARE Principles for Indigenous Data Governance? Are they able to use the framework to help prepare for or inform their work relating to Indigenous collections?
CARE Engagement example discussed during the roundtable:
The British Library (BL) has rolled out their Anti-Racism Project (ARP) which involved the formation of working groups drawn from members of staff from all levels, across all areas of the Library with the aim of enacting a generational shift to support the BL to become more representative and diverse organisation. The recommendations from the working groups formed the basis of a Race Equality Action Plan (REAP).
The REAP project was tasked with benchmarking the Library’s current performance across a range of key areas affecting racial inequality; including HR policies, Global Majority representation within senior management roles, inclusivity of exhibitions, events and public programmes and interpretation of the Library’s collections.
Challenges
- There are many different types of communities and although some sector organisations want to centre marginalised voices and change practices, some organisations do not know how to listen.
- Ethical issues can arise when working with a diverse range of materials including images, texts, audio visual materials, that are either born digital or digital surrogates. This includes materials that may be considered for repatriation or the involvement of the communities of origin.
- There can be a lack of awareness of what is in the collections, or if the language being used to describe collections is offensive. For example, conflicts between communities in how language is used to catalogue objects that can result in a recognised Anglophone bias.
Opportunities
- Organisations can develop skills to navigate conflicting views about collections. Managing objects on a case by case, contextual basis is a great way to improve the organisation’s skills and knowledge.
- Building on the British Library’s experience and training in the CARE principles, curators and universities can learn to design projects so that the CARE principles can be embedded from the outset.
Second provocation: How might the cultural heritage sector reimagine its practices to align with Walters’ vision of an Indigenous data voice? In what ways could principles of mutual benefit, reciprocity, and relational accountability enhance institutional and sector-wide practices to foster more ethical and equitable approaches?
Authority and collaboration
Who has authority? One individual? A group of key figures, or the whole community? What does sufficient collaboration look like? Careful stakeholder mapping will be important to ensure heritage organisations are speaking to the right people.
Assumptions and biases
Heritage organisations can be biased in response to the voices of communities of origin and can also have an assumption of what they will want from digital practice. There needs to be a willingness to understand the history of organisations and collections. Research within universities requires an ethical review. Often this may be a self-assessment, which can mean the ethics are only considered by the researcher themselves. How can ethical reviews ensure that bias is addressed?
Open licensing and ethics
Guidance is needed on the ethical use of open licenses. How do we ensure the legacy of the output is ethical, when it is being used, shared or reinterpreted?
Challenges
- The issues raised are complicated – because large collections may lack good provenance information which can lead to additional challenges of identification, cataloguing and contextualisation. AI might help with the work of building a decent knowledge base.
- Communicating with stakeholders can take time and resources – and consultation is a skilled action requiring a deep understanding of cultural sensitivities.
- Big institutions have the resources and technology to do this work and will often impose the same standards of partnership working on communities of origin and their representative groups – despite these participants not having the same resources or expertise.
Opportunities
- The development of codes of ethics for collections and museums, mock trials to see if practices are against code of ethics or against the law. This might be modelled on the ethics approval process for higher education research.
- Working forwards not backwards – embedding inclusive practice and creating an inclusive archive to counteract how institutions are embedded in colonialist and other power practices.
- Need to rectify mistakes of the past where cultures have been obliterated or distorted through attempts to story tell by outsiders. Our collections are not UK focused, but connected to people around the globe. Our curatorial actions have impacts internationally.
- Archivists can act as community facilitators to build trust and recognise and work with trauma, if they are properly trained and supported.
- The FAIR principles are already widely used across disciplines. There is a clear opportunity to promote the CARE principles to further improve collections practice.
Keynote Speaker: Kelly Foster
Kelly Foster addressed how power dynamics in open licensing disproportionately affect marginalized communities. She highlighted the importance of reparative approaches and how the sector can avoid perpetuating inequities in digital spaces. She also discussed the importance of Black Digital Practice in redressing those inequalities.
Black Digital Practice was coined by Jessica Marie Johnson in her paper “Markup Bodies: Black [Life] Studies and Slavery [Death] Studies at the Digital Crossroads”: “[Black Digital Practice identifies the people who…] hack their way into systems (whether modernity, science, or the West), thus living where they were ‘never meant to survive'”. Black Digital Practice refers to the ways in which Black communities engage with digital technologies, leveraging them for creative expression, cultural preservation, and social justice. This concept encompasses the use of digital tools to amplify Black voices, promote activism, and create spaces for storytelling and connection. It emphasises the importance of representation, community-building, and the reclamation of digital spaces, highlighting both the challenges faced by Black individuals in the digital realm and the innovative strategies they employ to navigate and transform these spaces. Ultimately, it underscores the intersection of technology, culture, and identity within the context of the Black experience. https://read.dukeupress.edu/social-text/article-abstract/36/4%20(137)/57/137032/Markup-BodiesBlack-Life-Studies-and-Slavery-Death
Third provocation: Can we talk about “Black data sovereignty” when there is no Black sovereignty? In the absence of a governing authority how can we think about ‘authority to control’?
Community Engagement Example Discussed During the roundtable:
A collection of Imperial era clocks from Beijing were on display in the Science Museum. In this case, the narrative was written by visitors and members of the Chinese community, in partnership with the China Exchange.
Challenges
- In large institutions, it is difficult to allow partnerships to actually lead work. Often there is push back or an expectation that the institution is the expert. Flexibility is essential, particularly when exploring best use of technology and data.
- Open Access licences can be individualist and used in ways that exclude or do not benefit the public and/or communities of origin.
Opportunities
- At a project level- at start of a project ask what the goals are, who is participating and who benefits?
- There is a significant need for greater diversity at governance level in cultural institutions, and appropriate support and resource for board members.
Fourth provocation: How might the UK’s heritage sector engage with Black Digital Practice and reparative approaches? What is the potential impact of digitalising the ‘deficit’ model present in open data?
Community engagement example discussed during the roundtable:
Hackney Archives has experience working with archives from the 1980’s that includes Black British community activism and responses to police brutality. It received funding from the National Archives to catalogue its collections, but Hackney Council Archives staff needed community engagement archive expertise as well as resources to respond ethically and sufficiently to the trauma caused in experiencing the materials in the archive. It was very difficult to recruit for the role and substantial resources are needed to manage and support community collections. Hackney Museum and Archives have a long historical commitment to represent minoritised racial communities in realising cultural rights, coming from broader anti racism strategies of the 1980’s policies of the Inner London Education Council and the Greater London Council. Time and training are needed to do this work in a very diverse and mobile borough. Payment for community expertise to work as a Steering Group, contributing knowledge and time is crucial for stability.
Challenges
- Telling accessible and participatory stories requires consistent, long term community engagement. Resource will be needed to establish inclusive and antiracist practices.
- Silo collections and siloed working; thousands of small organisations are run by small groups or volunteers – so training and learning is limited, and resources are not shared.
Opportunities
- There are real opportunities for Global Majority archives to become caretakers of their own culture given proper resource
- Reciprocal opportunities through more support for archives that focus on diverse community groups. Large institutions could engage with smaller archives to learn from them and in return share infrastructures, resources and frameworks.
Conclusion and actions
Key takeaways
Session 1: AI
Participants noted that they are anxious about the impact of AI on the cultural heritage sector, in particular regarding copyright, income, and personal data, and would benefit from more accessible support in understanding the current legal, ethical and technical issues.
Session 2: Indigenous sovereignty and anti-racism
Attendees expressed that although their collections may not include materials created by indigenous peoples, most UK collections contain materials from other countries and communities of origin. There are important lessons to be learnt in the UK cultural sector from the CARE Principles.
Reflection on outcomes
The roundtable highlighted the significant changes and challenges facing the heritage sector in relation to AI advancements and the work of Global Majority activists, academics and archivists. Skills gap, resource pressures and costs were identified as reasons opportunities were being missed for the sector to evolve to support innovative practices and publics and user groups. Open licencing provides significant opportunities for collaborative, inclusive and ethical practice across heritage, although public interests and resources could be much better protected in relation to AI development and regulation. Heritage institutions can benefit from ethical and inclusive practices, ensuring that all communities, particularly Indigenous and marginalised groups, have a voice in how their cultural heritage is represented and shared, and to ensure that open sharing isn’t inappropriate.
Actions
Key areas identified for further exploration by heritage organisations include:
- Development of clearer organisational policies and practices around AI use for UK cultural sector workers
- Better open licencing and copyright information, along with information about the current rules about data harvesting and reuse in the UK and internationally. Many participants believed incorrectly that open licence agreements were being used as the basis for AI scraping.
- Working collaboratively across the sector, and with other sectors (such as Higher Education), to develop shared positions on commercial use, public access and remuneration of cultural assets by AI companies.
- Greater discussion of the importance of a public digital commons and public digital goods and services for ensuring the sustainability and accessibility of UK heritage
- Resources for the sector to better understand the ethical value and limits of using open licences to share materials, that engage with community led frameworks including the CARE principles.
- Further exploration of the importance of Black Digital Practice in relation to the care and governance of digital collections, including open licencing and access.
- Access to funding to develop pilots and initial innovators in the AI space that are relevant to current and new UK collections.
- Ways for the cultural sector to understand and calculate the environmental costs of AI use.
- Cultural sector consultation by UK governments about the cultural value of regulatory and legislative approaches to AI and open licencing of publicly funded assets
- Development of practices to ensure better inclusion of historically marginalised and excluded communities in the use and display of culturally marginalised content, including at the level of decision-making, governance and management of collections, including providing information for community organisations to contact cultural institutions.
Vision for the Future
The roundtables underscored the need for leadership and innovation to shape the future of UK collections by heritage organisations. By engaging with new technologies, fostering equity, and ensuring ethical stewardship, the sector can continue to evolve in ways that honour both the past and the future of open knowledge.
Speaker biographies
[Photo by Laura Cameron 2021]
Except where noted and excluding company and organisation logos this work is shared under a Creative Commons Attribution 4.0 (CC BY 4.0) Licence
Please attribute as: "Changes and Challenges in Heritage and Open Knowledge (2024) by Naomi Korn , Sean Waterman and Naomi Korn Associates supported by The National Lottery Heritage Fund, licensed under CC BY 4.0
