Universities need to help improve AI-driven internet searches

It is vital for librarians to work proactively with tech firms to address the limitations and biases of the likes of ChatGPT, says Leo S. Lo

May 7, 2023
A man looks through binoculars, symbolising internet searching
Source: iStock

With the arrival of Microsoft’s GPT-4-powered Bing search engine, the information discovery process is on the cusp of major disruption.

For the past two decades, the keyword-based Google search model has been dominant, with Google itself holding a 93 per cent share of the search-engine market. However, the unprecedented impact of ChatGPT, which acquired more than 100 million users only two months after its launch, and Microsoft’s swift integration of the technology into its search engine could quickly turn that on its head. So, if librarians are to remain relevant, we need to swiftly integrate AI technologies into our own practices and confront the technology’s ethical and technical limitations.

It remains to be seen how the battle of search engines will play out, but there is no question that generative AI will transform the traditional keyword-based model Google uses, which also has a significant impact on the way library databases and catalogues are designed.

The use of keyword search queries can be an efficient and accessible way to look for information, particularly when the user knows precisely what they are looking for. Yet, because of its reliance on the rating and ranking of the diverse sources, one significant drawback of the approach is that it can be challenging to identify trustworthy and pertinent sources of information. This is why information literacy is essential – and librarians play a crucial role in teaching this. Our expertise in information organisation, retrieval and evaluation makes us well-suited for guiding users through the information-seeking process.

The new ChatGPT-powered Bing search engine uses natural language processing to generate personalised responses geared to the user’s needs and interests. By answering their questions directly, it can save them time, removing the need for them to look through pages and pages of website links. The generative AI model, however, has a number of drawbacks that must be considered.

First, the algorithms that drive it are complex and opaque, which makes it difficult to understand how it chooses what data to offer users. This raises questions about the accuracy and dependability of AI-generated search results.

Concerns about the ethical implications and limitations of AI, such as bias, discrimination and data privacy, must be also addressed to ensure the responsible and effective use of these technologies.

And the fact that the quality of the outcomes is only as good as the data that the AI algorithms are trained on presents another key problem. The search results might be inaccurate or incomplete if the data used to train the algorithms is inaccurate or incomplete.

It is crucial for library leaders and researchers to be aware of these changes and to act quickly. After all, academic librarians will need to train and educate users on how to use generative AI search engines effectively. This will entail teaching users how to ask effective questions and how to evaluate the AI’s responses. Keyword searches will evolve into what is known as prompting or prompt engineering, with librarians playing an important role in directing users to reliable and relevant sources of information.

Librarians will also be crucial in ensuring that AI algorithms are accurate, dependable and transparent. They should begin experimenting now with generative AI search engines and other AI tools to determine their potential advantages and disadvantages. For instance, they could run the algorithms through a variety of datasets and scenarios to find any problems or limits.

Librarians can also work to make sure that AI algorithms are as clear as possible so that users can see how they work and how they came up with their answers. Librarians must therefore collaborate with developers and vendors, providing feedback on how these tools can be better suited to academic libraries and their users’ needs.

University leaders also need to step up to the plate. The appropriate and effective use of AI in academic libraries is critical for higher education institutions’ performance and competitiveness in the digital age. Investing in the appropriate and effective use of AI can directly improve student achievement and lead to more fruitful research and learning outcomes, with libraries playing a crucial role in fostering both.

So this is an opportunity for higher education leaders to be proactive. By supporting and encouraging their library’s use of AI technologies and services, they can ensure that their institutions remain competitive in the ever-changing higher education landscape.

Leo S. Lo is dean and professor in the College of University Libraries and Learning Sciences, University of New Mexico.

Register to continue

Why register?

  • Registration is free and only takes a moment
  • Once registered, you can read 3 articles a month
  • Sign up for our newsletter
Register
Please Login or Register to read this article.

Related articles

The AI chatbot may soon kill the undergraduate essay, but its transformation of research could be equally seismic. Jack Grove examines how ChatGPT is already disrupting scholarly practices and where the technology may eventually take researchers – for good or ill

16 March

Sponsored