This Working Paper reports pilot empirical work examining a sample of African NLP research papers published between 2019 and 2024 along these dimensions: the type of resource created, the data sources used to create them, the licences attached, the funders and supporters behind the work, and the repositories on which they appear. It presents preliminary findings, notably that the data sources behind African NLP are predominantly copyright-protected materials from news media and personal information; that the resources produced are themselves frequently released without clear or explicit licences; and that an overwhelming number of hosting and funding and investments for this work is from outside the continent.

More details and abstract.

 

This Report examines the interaction between AI development and IP governance across 10 African countries: Cameroon, Ghana, Kenya, Namibia, Nigeria, Rwanda, South Africa, Tanzania, Uganda and Zimbabwe. It analyses how existing copyright, trade mark and patent regimes intersect with the AI lifecycle; identifies emerging IP governance tensions within the AI ecosystem, and evaluates how national AI strategies and policy frameworks across Africa are beginning to address those challenges. It explores the extent to which existing IP laws are clear and adequate to address the use and creation of IP-protected materials in the development and deployment of AI.

More details and abstract.

Front cover of report

 

Discussions around artificial intelligence (AI), data science, machine learning, text and data mining (TDM), etc. have made things so that everyone most are aware that access to data is imperative for these activities. In this regard, the role of legal frameworks such as copyright law, data protection laws and contract law in regulating and structuring data access is significant. The Data Science for Social Impact, a research lab at the University of Pretoria invited Dr Chijioke Okorie to speak at its seminar series and she used the opportunity to explore, amongst other things, the role that these legal frameworks play in facilitating or restricting data access within the African continent.

Full report of Dr Okorie’s presentation is available on The IPKat Blog and the video recording may be assessed on YouTube.

This was a presentation by Dr Chijioke Okorie at the 2023 Flexible Futures Conference held in August 2023. The presentation was delivered under the sub-theme: Implementing authentic teaching, learning, and assessment approaches in an AI dominant future.

The abstract is reproduced below:

In higher education, traditional forms of assessment are still dominant, and advocates of these assessment methods argue that they can effectively reduce plagiarism, are dependable, and are straightforward to create, grade, and administer. However, one limitation of these assessments is that they are typically closed book, preventing students from using reference materials and therefore leading to a greater emphasis on the product rather than the learning process itself. Moreover, poorly designed tests and examinations may assess knowledge that is disconnected from real-life situations. This is even more challenging in a field such as intellectual property law where real-life application of knowledge and learning is important to develop future-ready lawyers. To address this issue, incorporating authentic assessments and taking cognisance of AI developments can prove to be a valuable strategy. Authentic assessments strive to replicate real-life tasks, fostering attributes like autonomy and motivation, which are highly valued in the legal profession.

In this study, we report on the experiences of implementing authentic assessments in the intellectual property law module at the University of Pretoria. The purpose was to establish students’ experiences in the assessments where they were required to design and create an online education resource to explain aspects of copyright law. Students were working in groups and were allowed to design the online resources of their own choice based on their assigned copyright law topic.

Apart from the creative and design elements upon which students were assessed on their ability to make design and creative selections that are appropriately aligned with their assigned topic, the assessment used in this study also had group reflection as a component. Each group was required to complete a survey to inter alia indicate copyright issues they encountered in using any third-party material in their design; a brief description of the process of creating the submitted work, what they learned from the work, contributions by each individual member of the group, etc. This survey was not graded but to ensure that students participated, no group will receive a mark for the assessed part if they did not participate in the survey.

The survey not only helped students to reflect on the practical and real application of copyright law to creative processes and creative materials; they were also able to see copyright concepts such as joint authorship; originality (i.e., through the requirement to describe the creative process of each group); licensing and, the value of the public domain (i.e., through the requirement to indicate issues with dealing with third-party material and how that influenced the choice of materials used); etc. While some students indicated that their group did not have to deal with any copyright issues in using any third-party material – audio, text, audio-visual and visual materials, etc. in creating the group work. Of the student groups that did have to deal with copyright issues, there were varied selection in terms of public domain and/or royalty-free materials that they opted for.

The study established the significant value of authentic assessments for law professionals, as they enable individuals to demonstrate their comprehension of legal concepts and principles, and effectively apply them to authentic scenarios. By engaging in authentic assessments, law professionals can develop and demonstrate important qualities such as autonomy and motivation. However, a future research direction would be to fully incorporate generative AI models into authentic assessments such as the one undertaken in the study described.

Conference book of abstracts is available here. Request full version of presentation here.

This paper written by Dr Chijioke Okorie as part of the Program on intellectual property and Information Justice’s right to research in international copyright law project, sets out the issues of copyright ownership and risk of copyright infringement liability raised by data science research use of data held by public bodies (in particular, public service broadcasters) in South Africa. Considering both the fair dealing exception in South Africa’s Copyright Act of 1978 and the proposed fair use provision in its Copyright Amendment Bill B13F-2017, the paper discusses these issues elaborating on the reasons why data science researchers in public research institutions should not require a copyright licence or be considered to be infringing copyright when they use copyright-protected materials held by public bodies for data science and artificial intelligence or machine learning research (henceforth, data science research). The paper also suggests that even where the outcomes/outputs of data science research are copyright-protected, they should be made available in an open and accessible manner with reasonable safeguards.

The paper is available open access.

This book chapter by Dr Chijioke Okorie forms part of an edited volume on Artificial Intelligence and the law in Africa.

The chapter is premised on the argument that the use of the IP system to take advantage of the opportunities, and overcome the challenges, of AI on the African continent should be based on the recognition of the contextual characteristics that exist in the continent as embodied in expressed development plans. The chapter takes a brief but closer look at South Africa’s NDP 2030, highlighting the role of AI in the actualisation of the development goals. More significantly, the chapter showcases the contextual characteristics of South Africa as a microcosm of the African continent to show how such context should necessitate a different AI IP policy. It concludes that the imperative for South Africa and other African countries is to facilitate AI applications across all sectors of the economy. Such AI applications in health, education, governance etc. will contribute to Africa’s development. The chapter also presents an overview of the varied uses of the IP systems (specifically copyright and patent laws) to promote creativity; inventiveness and innovation. On this basis, the chapter supports the argument that the grant or conferment of IPRs is not the only way to facilitate investment and innovation in the AI space or to harness the opportunities of AI or address the challenges posed by AI. Rather, the IP system also offers scope delineation and exceptions to IP protection that can also (and perhaps better) facilitate investment and innovation in the AI space. The chapter concludes by proposing strategies within the IP systems that can enable Africa to take advantage of the opportunities, and overcome the challenges, of AI on the continent.

The chapter is available open access.

Dr Chijioke Okorie facilitated a workshop and dialogue session at the 3rd Edition of Hundzula: Natural Language Processing and Linguistics Retreat held at Nelson Mandela University, South Africa. The session titled Licensing corpora for natural language processing [#AfricaNLP] was made possible through financial support from Meta.

Hundzula: Natural Language Processing and Linguistics retreat, brings together researchers working in Natural Language Processing and Linguistics to learn, teach and assist each other.

Reflections on the retreat and the workshop specifically to follow shortly!