Report on the webinar “AI waits for no-one” (Tuesday 23rd April), by James Andrews

 

GIG recently joined forces with colleagues in CILIP Knowledge & Information Management Group for the second in K&IM series of Artificial Intelligence webinars. The recording of the event has been uploaded to YouTube. Here, in an article originally written for the K&IM Group’s newsletter, James Andrews reflects on hearing from two speakers working in the information field with different backgrounds but who were both able to use their knowledge of the technology to shed light on its fast-changing contours.

As the seemingly endless news stories about AI and its impact on jobs and society continue to proliferate, the CILIP Knowledge & Information Group and Government Information Group joint webinar on AI in our profession was a timely opportunity to take stock. With two distinguished speakers, Dr Andrew Cox (University of Sheffield) and Dr Kelcey Swain (Cabinet Office), offering perspectives on AI application in knowledge and information work, this was a chance to listen, learn, and question.

 

Echoes of Edwin Starr’s iconic protest song of the late ‘60s instantly sprang to mind - would we get an answer to the question  ‘AI, [huh...] what is it good for?’

 

Andrew Cox took a broad overview of AI use suggesting that KIM professionals should welcome its potential. Whether generative or otherwise, AI is enhancing access to information through capabilities in translation, transcription, and captioning. Tools such as Microsoft’s Copilot, though imperfect, are powerful aids for knowledge workers when drafting documents, tweaking communications, and performing basic spreadsheet functions.

 

New challenges are emerging from AI use, such as the need for effective prompt engineering and addressing potential biases in large language models. Andrew proposed establishing 'AI literacy' training, potentially as an extension of existing information literacy efforts. This concept has great merit and the need to develop and embrace such training will likely become increasingly critical.

 

If you are concerned about AI impact on KIM work, take heart in Andrew’s conclusion - KM skills, ethics, and values have never been more vital.

 

Kelcey Swain’s presentation focused on a practical AI use case. He provided an overview of his work on using linked data sets to develop a tool that identifies names, places and organizations within Cabinet Office documents. Using tools, like the spaCy Python library, he is creating a product to assist with this task, though it has limitations. Disambiguation of terms is a key challenge for ensuring relevant documents are highlighted based on context. For example, Cabinet papers on defence spending referencing “Michael” in 2015 likely refer to Michael Fallon, whereas “Michael” in 1990s cabinet papers probably refer to Michael Heseltine. To improve the tool’s performance, Dr Swain is training the AI model on context-specific datasets for additional decision support.

 

The Cabinet Office has a vast digital archive and as content continues to proliferate, searching for a needle in an ever-expanding haystack poses greater challenges. The promise of AI models to support information retrieval for internal queries and Freedom of Information requests is clear.

 

Returning to Edwin Starr - following an hour with the two doctors, am I any clearer on what AI is good for? It turns out quite a lot, though with the caveat that it is far from perfect and will pose both new and familiar challenges. As with any technological shift, organisations and societies will need to adapt accordingly and I am confident that KIM professionals are well-positioned to help develop AI tools and address emerging challenges. Watch out for the opportunities and embrace them when you can!

Thanks to James for sharing his thoughts and for members of the CILIP K&IM Committee for agreeing for them to be published on our blog.