The impact of AI in libraries… so far

As the initial hype around AI begins to die down and people everywhere start to engage meaningfully with the technology, questions are beginning to be asked about what the latest generations of AI are good for and where they might be useful.

The perhaps predictable consensus that is emerging is that AI is very useful for automating routine tasks, from producing chipper and upbeat text for social media posts to identifying likely subject headings that appear in our catalogue records and that you can click on to find more books in our collections on particular topics covered by books that you find. People have used AI for years to help them improve the flow and style of their writing, helping writers sound friendly and succinct without compromising their own voice. Unsurprisingly, this continues to be where most enthusiasm is being shown.

AI is also showing promise for solving the typical office problems of organisation and planning, from Goblin Tools showing how to structure tasks efficiently to leading collaboration tools from Microsoft’s in-built AI assistant Copilot to the new AI-augmented search and summary service now offered by Slack attempt to find and retrieve information from loosely structured conversations and documents in response to user queries. It serves this function well since generative AI is becoming good at summarising what has already been said, particularly for small datasets.

Few people were sufficiently disciplined in the ways they organised information to allow them to plan ahead and structure information in easily understandable ways so they could find the information again easily. Generative AI tools appear to be the successors to simple algorithmic search engines, allowing intranets to be searched, queried, and now summarised with the same ease as an internet search. The latest iterations of AI are also reportedly paraphrasing sources rather than copying and pasting text, although the jury appears still to be out on just how accurately they represent the original content.

So, while the arts are weathering an assault of original work being scraped by generative AI without the creators’ consent, AI workplace solutions to date appear to be a little more tame and ethical. The terms and conditions of licences, particularly what rights AI has to store and analyse the content it is sifting through has attracted considerable attention and is probably the biggest concern for businesses who do not want to wake up one day to realise they have shared all their deepest corporate secrets with AI giants. How roles and practices adapt as people begin to find the most effective and efficient ways to use these tools, we will just have to watch and see but so far, they are continuing usefully to save time without challenging the original thought and decision making reserved for humans.