Navigating the new era of search: Implications for library and information professionals

A reliable and effective natural language search option has been a long time coming. Although this functionality has existed for a while, many library and information professionals have stuck with the traditional boolean search. What can I say? We like to be in control of our searches and understand why a database or search engine returns the results they do. 

Therefore, what do the latest developments in GenAI as a research tool mean to us as information people? The rise of GenAI and the evolving functionalities of search engines like Google and Bing are reshaping how we access and interrogate information. These tools offer natural language processing capabilities, making information access more intuitive but also raising concerns about accuracy and reliability. 

This blog post explores these changes and their implications for library and information professionals - and explores a couple of interesting articles that have appeared recently.

The (d)evolving role of search engines

This post was inspired by the blog post by RIPS Law Librarian. It raises some interesting issues about the diminishing effectiveness of Google search, especially in the context of legal research. It highlights a study which suggests that the prevalence of affiliate marketing and SEO-driven content in search engine results, is obscuring useful and unbiased information. How do we overcome this? 

By being sceptical about everything you find online. Some things never change! In the above post, one of their recommendations to end users is to bypass the search engines and start with a trusted website and I agree. Firms, organisations, and institutions spend a lot of money on paywalled information resources so you want those to be the springboard for everyone’s research. 

However, these specialist databases - for the most part - require the end user to have search expertise. Wouldn’t it be great if people could get what they needed by simply having a conversation with a database? ChatGPT and other forms of GenAI allow us to do just that. We can start with a question and then refine our responses until we get the information we need.

How are publishers meeting the needs of end users?

We want our end users to find reliable and relevant information. We have already seen exciting functionality being rolled out by large legal publishers (Lexis+AI and vLex’s Vincent AI to name just two). Vable has been exploring GenAI options and we are looking forward to sharing more about this when we are ready. There are many reasons why we should be excited:

• End users deserve a simple interface: There's a growing expectation among users for more intuitive and user-friendly search experiences. When end users read about GenAI tools like ChatGPT, which offer conversational and natural language interfaces, they will expect similar functionalities in professional paywalled databases.
  
  
• End users deserve better (whatever that means to them) search results. Natural language processing (NLP) capabilities, a cornerstone of GenAI, can significantly enhance the search functionality by understanding and interpreting complex queries. 
  
  
• Technological feasibility: The advancement in AI and NLP technologies makes it more feasible for database providers to integrate these features. The cost and complexity of implementing such technologies are decreasing, making it a viable option for many providers - and more cost-effective for end users.
  
  
• Enhancing content value: By making their content more accessible and easier to search, paywalled databases can enhance the perceived value of their offerings. Easier access to information could lead to increased usage and citation, benefiting both the database providers and the content creators - and ultimately, end users.
  
  

A review of the recent Legal Information Management article

Once you are aware of a particular issue, for example, the use of GenAI in legal research, you start seeing articles everywhere! Happily for me, the well-researched and timely article, "Is ChatGPT Any Good at Legal Research – and Should We be Wary or Supportive of it?" appeared in my current awareness today. Greg Bennett's article critically examines ChatGPT's effectiveness in legal research.

Taking advantage of the premium version of ChatGPT with the 'KeyMate.AI Search' plugin, he tests its capabilities on various legal research tasks, including interpreting legislation and referencing in OSCOLA format. He concludes that while ChatGPT demonstrates proficiency in some areas, such as accurately detailing legislative amendments, it struggles with complex legal queries and precise referencing. This is a well-known limitation.

The article also addresses ChatGPT's "hallucinations," where it generates plausible but incorrect information, and its potential to access content behind paywalls, raising concerns about the future of legal research and the role of law librarians. The discussion mirrors concerns raised in the RIPS post about the declining usefulness of Google search for legal research, emphasising the need for critical evaluation of AI tools and the continued importance of human expertise in legal information management.

The importance of human expertise

GenAI is an intermediary between search engine content and the human interrogator. The problem remains that we need the quality of information to be right from the start - I hope that we have learned to be critically aware of search engine results and will continue to apply this scepticism to these emerging intermediaries! Why not think of this tech as an opportunity for professional development.

Let’s keep the following in mind:

• Use clear prompts: I’ve already written about the need for the right prompts. Outline your required tone and always provide context for your query. Even when working with paywalled GenAI functionality, the more information you provide, the better the response will be. Publishers will provide advice on using their tools so make use of their guidance, where possible. Keep in mind, as the tech evolves, your prompts will need to evolve. 
  
  
• Check everything: Don’t accept anything as immediately factual from your initial question, and remember if it looks suspicious in any way, challenge it! ChatGPT gave me some statistics on the number of information people were using GenAI for legal research. I immediately asked it where it found them. Its response - 'the statistics were illustrative only, you must go and find a report for the relevant figures'. Indeed! Always refer back to trusted sources or colleagues. 
  
  
• Critical evaluation: This intermediary role of GenAI underscores the importance of critically evaluating the information provided by these tools - especially when we consider from where they are getting the information. Users should be aware of the potential for inaccuracies and biases and may need to verify information through additional sources or direct searches when accuracy is crucial​​​​. Use your critical skills!

While GenAI tools offer an interactive way of accessing and engaging with information, they are intermediaries that reflect the strengths and weaknesses of their source data. The quality of the output is contingent on the quality of the input, and this relationship highlights the ongoing need for improvements in data quality and reliability across digital information sources. They also require end users to be knowledgeable, critical, and aware of the pitfalls.