Libraries around the world and in South Africa are incorporating artificial intelligence (AI) innovations as a complementary tool to the services provided at libraries to better serve users.

Librarian tasks such as cataloguing materials to make them more accessible online, issuing and returning library books, and renewing those books are now AI-driven.

However, experts say that AI and robotics will not cause library employees to lose their jobs, but staff will be trained for more professional roles, as the field transitions to AI and robotics.

These details were revealed during the 5th IFLA Librarians Symposium, held by the University of KwaZulu-Natal, in partnership with the International Federation of Library Associations (IFLA) Information Technology Section, and IFLA Artificial Intelligence (AI) Special Interest Group (SIG), in Pietermaritzburg, KwaZulu-Natal, recently.

This year’s theme was Artificial Intelligence and Robots Transforming Libraries: Opportunities and Challenges, with a focus on the Southern African region.

Experts from around the world in fields such as AI, robotics, and Library Information Services shared best practices in generative AI, the importance of ethical AI deployment in supporting research, teaching, and library operations, during the symposium.

The responsible use of AI in education involves using AI tools to enhance learning, support educators, and promote student well-being while respecting ethics, data privacy, and academic integrity.

Professor Anil Chaturgoon, the acting deputy vice chancellor for Research and Innovation at UKZN, reflected on the role that libraries have played throughout history and described them as a timeless hub of knowledge and innovation.

He said from the ancient scrolls of Alexandria to the digital archives of today, libraries have preserved humanity’s collective wisdom and made it accessible to all.

About the new era, he said AI is now a transformative force rather than a distant dream or niche academic pursuit.

“From healthcare and education to entertainment, transportation, and beyond, AI is changing how we work, how we communicate, and how we solve some of the world’s most pressing problems.

“With great power and advanced technology comes great responsibility as we develop and deploy AI technologies. We must ask ourselves, how do we ensure that the tools are ethical, inclusive, and beneficial to all?” Chaturgoon said. 

He explained that these are not just technical questions, but deeply human questions that require diverse voices and perspectives.

Santiago Villegas-Ceballos, an expert in engineering, librarianship, marketing, and innovation from Colombia, north of South America, who uses his skills to transform libraries at universities around the world, told delegates that one of the dilemmas “we” have with artificial intelligence is the tension between representation and appropriation.

Villegas-Ceballos highlighted that on a large scale, AI systems are conceived, trained, and governed in the Global North, on data sets that under-represent Southern languages, bodies, and worldviews. 

“The result is what scholars now call digital or data colonialism. Bias in training data systematically filters out our ascents, our stories, and our epistemologies, producing tools that speak of humanity while knowing a little about the majority of the world.

“Just as music has crossed oceans, and high validity has birthed new genres, we must learn to hack AI to recontextualise with our languages, our ethics, and our social needs, so that technology becomes a polyphonic chorus rather than a single flattened melody,” Villegas-Ceballos said.

Through a few search exercises on ChatGPT, he demonstrated that AI does not learn information in real-time; rather, it outputs what it has been trained on.

However, algorithms in social networks like TikTok and Instagram are in real-time learning about “us” humans.

He shared tools such as archive.org, which will land you on the Wayback Machine, an internet history archive site, which contains all facts about the information on the internet, from when countries first got the internet, the search engines they used, etc. 

These would help correct false information and can be input into AI to correct misinformation.

Villegas-Ceballos said just like AI, robots are programmed for a specific task they are made for, and fitted with infrared sensors to see what is around them, among others.

He encouraged delegates to use open-source (free) platforms such as Hugging Face to create their own chatbot assistants.

Thought leaders also shared how students and academics can critically evaluate AI-generated content so that academic integrity is maintained.

Professor Donrich Thaldar, from the University of KwaZulu-Natal’s Law School, said one has to know their field of study very well. 

“To critically evaluate AI output, you have to use the same method as you rate what any other human has to say on anything in your field. Verify the various statements that the generative AI has come up with to ensure they are factually correct. 

“You must always know that any tutor, whether it’s a human or a gen AI, can always make mistakes. So, it is always good to take the same statement to another generative AI tool or look it up on Wikipedia.” 

Even in school or university textbooks, it’s not always 100% correct, he said.

May Chang, an Information Technology director at the University of Cincinnati Libraries in the United States, said students and academics must verify all references, citations, and statistics to ensure accuracy.

“They must check for inconsistencies and inaccuracies in the text, even though it may sound confident. Be aware that potential bias in the output of AI-generated content is only as good as the data it was trained on. “Garbage in, garbage out” also applies here.

“Tools for creating or detecting AI-generated content are not fail-proof, and the human-in-the-loop and critical thinking skills are still necessary,” Chang said. 

!function(r,u,m,b,l,e)(window, document, “script”, “Rumble”);

Rumble(“play”, “video”:”v6t73sx”,”div”:”rumble_v6t73sx”);

At the symposium, UKZN showcased its first robot Librarian, named Thola, which is multilingual and designed to integrate with library databases.

UKZN Director of Library Services, Dr Nonhlanhla Ngcobo, said the university has taken a bold stride in the AI and robotic space, through its locally developed robot, which will be used to streamline circulation functions and improve user experiences.

UKZN is the second institution in South Africa to launch a locally developed robot librarian, following the North-West University.

Thola uses ChatGPT to communicate, has two digital ‘eyes’, and its chest has a tablet for conversational interaction. It can introduce itself, react to both pre-programmed and impromptu questions, including how to renew books or library cards, and library operating hours. 

It is fitted with wheels, but is unable to move on its own as the final touches are being completed.

Thola’s developer, Trevor Lorimer, was fascinated by its capability to take the core of answers provided in a standard Q&A session and provide them back to the audience in its own words.

Thola will be stationed at the UKZN medical school library, where further developments will be monitored and encouraged.

She said the UKZN library services will not introduce too many technologies at once. 

“We are focusing on one task at a time, and that’s also going to help our staff to familiarise themselves with the technology, to get used to the change before we can introduce another one. 

“We are going to start with our circulation functions done at our issue desks, issuing, returning, and renewing of library books. The focus will be on these first until we are confident that our staff have come to grips with the change. This will be done collectively with staff. We will then explore the next set of tasks that can be driven into AI,” Ngcobo said.

gcwalisile.khanyile@inl.co.za

undefined



Source link