Like it or not, GenAI is here to stay. After two and a half years of experimentation, we’re now well-acquainted with its vulnerabilities – hallucinations, ethical concerns, and accuracy issues among them. My LinkedIn feed is full of “end of the world as we know it” predictions but we’ve been here before with new technology.
Remember Google in 1998? Library and information pros were the first to recognise both potential and pitfalls. We took the lead with the testing, researching, and training to our end-users (and parents!), so people could navigate it with confidence. The same is happening with GenAI. So, what are legal information teams doing with it?
Most professionals are using it, or at least trying it, in their day-to-day work. But many people aren’t getting optimal or even accurate results. In law and other high-stakes sectors, that’s not just inconvenient; it’s dangerous. Courtrooms and regulatory environments demand precision. Fabricated case reports or inaccurate company data simply won’t do.
That’s where information professionals come in – again. Across the sector, we’re identifying, testing, and deploying GenAI in ways that are safe, useful, and relevant. In their BIALL session, Ashurst’s Lisa Paul and Susan Lamond explored whether GenAI helps or hinders in their work!
Budget time is a busy time with many hundreds of pages to process and summarise. The Ashurst team explored whether GenAI could help get the information disseminated more efficiently.
They tested CoPilot, ChatGPT, and Perplexity. They uploaded the documents into each and prompted the tool to extract and structure data such as department names, initiatives, values, timeframes, and more, into a table.
What happened? The tools were fast but flawed. Financial figures were off, initiatives were misclassified, and the overall accuracy wasn’t trustworthy.
Outcome: With careful prompt refinement, they improved the results but still had to manually verify each entry. GenAI saved time, but didn’t eliminate the need for oversight.
Taxi reports offer a snapshot of a company: leadership, financials, market footprint. Extracting this manually can be slow and painstaking.
With the same tools as before, the team designed a prompt asking them to scan official sources like websites and filings. They wanted the data tabulated and structured.
What happened?
Outcome: Surprisingly successful and effective. With checks in place, this task could become a self-service option within the library function, freeing up staff for higher-value work.
Sifting through and analysing lengthy reports can take time. With the right prompt, could GenAI surface a company’s strategic priorities? They tested “What is the business strategy of XYZ company? Please display this in a table using up to 6 themes.”
What happened? Once again, as with the Taxi Report, the results were presented as specified and looked satisfactory. The team drilled down into each section, refining the prompt to get the information they needed. For example, global comparisons, financials, achievements, and future direction.
Outcome: Another promising use case. Well-crafted prompts yielded useful insights. Again, with the caveat that content must be checked. Time saved here can be reinvested in deeper research support.
Current awareness is a time consuming task. The team wanted to know whether GenAI could track what competitor firms are publishing.
What happened?
Outcome: This use case fell short. GenAI struggled to aggregate and summarise competitor thought leadership effectively. Current awareness is still a specialist task which GenAI cannot yet do.
The Ashursts team had access to Harvey AI, which they were keen to test. This tool includes access to SEC data, providing useful company insights. Their initial prompt was, “Please list the top 10 tech companies.”
What happened? The list was successful and they went on to create further prompts to extract further information.
Outcome: Overall, this was a successful use case. Once the prompt was refined, the tool generated a useful summary of companies. Again, human verification was key.
Becoming proficient in GenAI tools doesn’t diminish the role of information professionals – it actually elevates it. These tools are powerful accelerators, but only when guided by human expertise. In legal and business settings, where precision matters, GenAI is not a solution in itself, t’s a starting point.
Across all these use cases, one message stands out: GenAI enhances efficiency but must be paired with critical oversight. Effective prompt writing, result validation, and source awareness are non-negotiable.
Information professionals are best positioned to take the lead. We know how to interrogate the sources, spot the gaps, and translate results into value. Just as we helped organisations navigate Google’s early days, we’re once again showing the way forward. Let’s keep testing, iterating, and sharing what works.
Try the powerful and easy-to-use KnowAll Matrix with no obligations, no commitment, no installation
 | Thank you for Signing Up |
Subscribe now to keep reading and get access to the full archive.
