Image courtesy of Hal Gatewood, Unsplash
Artificial Intelligence (AI) has assumed a leadership role in discussions about education. Neil Butcher’s blog on AI’s false promises in education lays out major issues and pinpoints the easy assumptions on AI’s importance to educational systems.[1]
Academic librarians can be at the forefront of helping users wade through the hyperbole, challenges, and benefits of AI. They have always been considered ‘the gatekeepers’ of knowledge. They can also open the door to better understanding of new technologies and concepts.
African libraries have been trailblazers in adopting technology. In the 1990s, librarians in countries like Malawi, Ghana, and Nigeria were responsible for email in their institutions and countries. In those days, there was no viable Internet, and email was based on store and forward messaging. Emails would be stored on a computer system and then forwarded to the organization responsible for distributing them.[2]
Christine Kisiedu, then university librarian of the University of Ghana wrote:[3]
"It is often said that once ICT [Information and Communications Technology] development gets underway, it is unstoppable. This was certainly the case in Ghana. When a workshop on Electronic Networking for West African Universities (sponsored by the Association of African Universities and the American Association for the Advancement of Science) took place in Accra in December 1993, Ghana was not counted among the Internet savvy countries of the sub-region. Two years later, in 1995, a nationwide store-and-forward e-mail system had been established, and the first professional ISP in Ghana appeared on the scene and introduced Internet access to an interested but cautious Ghanaian public. In another year, development and patronage had reached such a level that Ghana could be said to be on the verge of an Internet explosion! Yet coming to grips with the new technology was not without its ups and downs.
…Learning to use the system proved more difficult than anticipated. I had just completed my second year as University Librarian at the Balme Library when we acquired e-mail…Nobody on the library staff had a clue as to how to operate this technology, much less how it worked. I submitted myself to a brief half hour’s explanatory session, which I must now publicly admit went in one ear and out the other. The consultant took too much for granted!"
Artificial Intelligence (AI) is the latest and trending ICT development, and again academic libraries are playing a leadership role.[4] The librarian of Northwestern University in the United States curates a page on the library platform on Using AI Tools in Your Research.[5] Other universities have done the same. A Google search for ‘library guides AI’ pulled up scores of university libraries offering AI guidance:
In a recent blog post on AI and educational technology, referred to above, Neil Butcher pointed to the disparities between what we know about AI and its impact on education. He emphasized that international groups like UNESCO are promoting AI’s success with little evidence to go on. Butcher wrote:[6]
"ChatGPT was only launched in November, 2022 but, by 2021, UNESCO had already launched a full publication entitled AI and Education that provides guidance to policy-makers. It seems a remarkable achievement that the world’s largest educational intergovernmental organization already felt confident enough in its understanding of the educational use of technologies about which we know so little and whose potential wider social impact is so little understood to offer unequivocal policy guidance on their use to governments. Having worked in educational policy for many years, my instinct is that it will be several years before we know enough about these issues to be clear about the policy implications."
The African Library and Information Associations and Institutions (AfLIA) and OER Africa decided to ask African academic librarians about their experiences and thoughts on AI. Nkem Osuigwe, AfLIA Director of Human Capacity Development and Training, put out a call on the AfLIA-OER Africa WhatsApp group. She wrote:[7]
"Have you ever noticed how technology seems to throw up opinions about the relevance of libraries?. Remember when the Internet went mainstream, and there were all sorts of permutations about the survival of libraries? Many asked if the Internet would replace libraries. When artificial intelligence became the new buzz concept, quite a number of librarians jumped in to talk about robots and what they can be used for in libraries.
Now, generative AI is the new rave. Young and old can do research on AI platforms, and within seconds, 'everything' is delivered, including references. Generative AI is creating art, writing essays, poems, short stories, drama, newspaper articles, and even social media posts.
Who needs libraries now, one may ask!. The verdict is not in yet. However, in Africa, we want to begin to tell our stories by ourselves about how libraries could use this technology so that we can 'own' the narrative about the tool.
• what impediments do we foresee in the advancement of AI?
• Can information literacy modules/classes and/or Use of Library courses introduce our users to 'ethical' AI?
• Can librarians assist in detecting when people are attributed for work by generative AI?
• Copyright and licensing issues are bound to crop up with the continued use of AI. Can African librarians help on creating awareness about this?
Generally, we’d love to hear about our interactions with AI as African librarian."
Librarian responses have been numerous and interesting. They are still coming in. Below are some examples of experiences reported:
The University of Ghana has revised its plagiarism policy to include use of AI:[8]
"Any employment of AI or associated technologies that compromises the authenticity of academic output will be deemed unacceptable,’ the document stated.
The University stresses originality in scholarly work, addressing AI's role in academic integrity.
The policy update underscores a commitment to ethical academic practices, leveraging technology while upholding original thought."
Researchers from the global South are accused of using AI if the language of their research papers and proposals use uncommon words like ‘delve’. One participant commented on vocabulary and AI:
"There was an uproar last week on Twitter. An American funder got a good project proposal, but ditched it because he saw the word 'delve' and thought the proposal was AI generated, because he doesn't write using these fancy words; only machines do.
Apparently, some words are already been termed AI generated words, and so AI detectors would look for such words and flag.
What happens when we use these words frequently in our write ups?"
The tweeter was Paul Graham, co-founder of Y-Combinator and this is what he wrote:[9]
Graham’s post has been widely controversial, with many researchers responding that they use these types of vocabularies. For many people for whom English is a second language or who normally use an extensive vocabulary, words like ‘delve’ are a part of their writing and not AI generated.
Another participant noted:
"Africans, mostly Nigerians, jumped on the Twitter (now X) thread to point out that we write that way. I actually do. It appears that AI is yet to learn the difference between English spoken or written as a first language and how those who got it as a second language communicate with it.
It is functional prejudice that has been taken on by AI. What do I mean? If you play word games online, especially those created by those who have English as their 1st language, you would be amazed at the words that pop up as correct options. This supposition is still being tested, though, but Africans use more of 'official' and 'bogus' words more than the owners of the language. AI may have picked up that trait from the owners of the language."
In general librarians agreed that neither Internet nor AI can replace the library and used Nile University in Egypt as an example:
"If the Internet could not replace libraries, I don't see how AI could take the place of a library. Remember, we provide many services, and for now, I have seen scholars use AI Mostly for research purposes. Any good researcher should know that in conducting research, you cannot depend on AI tools alone; you have to consult other information resources. Just like what Nile University library is undergoing, librarians should see the coming of AI to our advantage and introduce newer services such as information literacy, how to detect plagiarism, and ethical issues among others."
More broadly and beyond AI, libraries play an essential role for all members of the academic community:
"Libraries need to consider transforming their spaces in research and learning Commons. The trend is catching on fast as some libraries also have worker and creator spaces. The client should feel as though they are in a one-stop shop where they can learn something and transform it into a prototype or sample product. Not only can we help job seekers, but we can hold meetings with guest authors. We can, if funds permit, partner a publisher to bring an author’s work to life in the library. My personal concern is the lack of partnerships and the slow evolution in our library spaces and the lack of recognition for those who spark creativity. I miss the time many people would speak about a book or library inspiring them. The voices are few, and a new crop that rejects the education system (inclusive of the library) is getting popular. Save for the places with a strong reading culture and push for education."
The WhatsApp AI discussions were informative and spanned the breadth of existing and possible interactions with their constituencies. AfLIA has made recommendations for libraries that want to engage more fully with their university communities on this topic:
Lack of knowledge about technology creates unfounded trepidation. AfLIA encourages African librarians to view AI as a tool for discovering more knowledge and reference sources for their user communities. Generative AI is a Large Language Model that is trained using large mostly open datasets. Viewing it through this prism will give librarians the understanding that generative AI recycles and collates already existing knowledge. That is an area in which librarians excel! African librarians have another advantage—the datasets AI uses do not include sufficient African source material, AI source materials therefore underrepresent African knowledge. Academic librarians understand the importance of appropriately indexing content, which is essential when considering datasets. African academic libraries and other institutions, such as archives, contain source material coming from Africa—both original sources and research content about African knowledge.
Innovation in information access is unstoppable. It is imperative that academic librarians adopt an open and adaptive attitude toward new developments. AI, including generative AI, just as any other tool or resource, has its limitations. But being apprehensive and unreceptive towards AI, probably with the intention of gatekeeping ‘a profession under threat’, is not sustainable – a lesson history has taught us time and again. In fact, that mindset numbs the innovative capacity of librarians and institutions to adapt and provide valued services to fill the gaps developed because of the limitations of generative AI. We must embrace it as an opportunity to evolve and better serve our patrons.
The time has come for African academic librarians to deepen engagement and collaboration across universities, both locally and internationally. The challenges faced by the profession can no longer be considered local, hence operating in silos is not viable. Through collaboration, we can collectively debate, brainstorm, exchange best practices, and share insights on what actions or decisions are being taken by academic institutions to address pressing issues like academic ethics, plagiarism, originality of thought, copyright, misinformation, disinformation etc. that have resurfaced, following the birth of AI resources like ChatGPT and Gemini.
The application of information ethics principles is important while engaging with generative AI. African librarians can lead the way in teaching and inculcating ethical skills in the use of generative AI through information literacy skills or use of library tutorials. The Wikipedia article on ethics and AI outlines key points of intersection:[10]
"The ethics of artificial intelligence covers a broad range of topics within the field that are considered to have particular ethical stakes. This includes algorithmic biases, fairness, automated decision-making, accountability, privacy, and regulation ."
AfLIA strongly encourages African librarians to understand and promote Openness as an avenue for making datasets available for training generative AI. In an article on AI and data, published by European Data, the authors make a number of points about the relationship between AI and data, including:[11]
"Open data and AI have the potential to support and enhance each other’s capabilities. On the one hand, open data can improve AI systems. In general, exposing AI systems to a larger volume and variety of data increases the chance of the system returning accurate and useful predictions. As such, open data can be a supply of large amounts of diverse information for AI systems. In this way, the availability of open data contributes to better performing AI."
If African librarians are to play a role in discussions about data, openness, and AI, they must understand the importance of data. The AfLIA collaboration with Wikimedia on data short courses[12] is important. But this work must delve more deeply into the significance of these relationships.
Library and information institutions may not directly dictate the data utilized in training various generative AI models to ensure a balanced perspective on African knowledge. However, it is vital for stakeholders to intensify advocacy so that African researchers, educators, librarians, and authors in general can embrace openness. Making African knowledge more accessible through open research, open data, and open educational resources can improve the availability of diverse and representative data used to train AI models.
Related articles:
[7] Nkem Osuigwe, AfLIA-OER Africa Training WhatsApp group