The excitement around the possibilities of Generative AI has continued throughout 2024 and is here to stay into 2025. We have shared here our experiences using this technology first hand and what is on our wish list for the year to come.
Generative AI is already proving to be a valuable tool. It can create meeting summaries, highlight action points, prepare briefing notes ahead of meetings, and extract data points from a large number of documents for review. It is making data review exercises viable, which would otherwise have been cost-prohibitive, and helping lawyers proactively manage their workload.
We have been intrigued by claims from providers of Generative AI tools of its drafting prowess however, in testing we have found that generic models have produced results that reflect the general narrative tone of internet content they have been trained on, rather than the technical drafting we expect to see in legal contracts. However, these tools are demonstrating their value in their ability to identify and extract specific drafting solutions across large scale data sets of precedents, reducing the need to scrabble around for past examples. This is an area vendors seem to be targeting, and we are paying close attention to advancements in this area.
The answer to the question "When will lawyers rely on Generative AI?" requires some thought on the barriers to adoption. Firstly, liability for the output is a factor. At present, products pitched to lawyers require there to be a human in the loop as vendors are not guaranteeing the accuracy of outputs. Although some vendors are starting to introduce insurance backed tools, currently these benefit the law firm and not the end user and so there is still risk for law firms in relying on unchecked outputs. Given AI’s continuing issues with “hallucinations” and propensity to state inventions with the same conviction as facts, it is difficult to see how law firms will be able rely on results if they are ultimately liable for the output. Secondly, the nature of errors presents a different risk profile. While overall accuracy levels of a Generative AI review vs a human review may be similar, people tend to make mistakes across a range of areas, whereas Generative AI tends to get the same thing wrong consistently. For example, we trialled a product that we considered to be 95% accurate at reviewing leases, but it incorrectly stated the rent figure nearly every time.
In the realm of Real Estate, Generative AI can struggle to piece together the various factors needed to report to clients. This means we need specialized tools to create a generative AI-generated report on, for example, a lease, which relies on the interconnection between various documents such as the lease itself, licences to assign, and rent review memoranda.
So, what's on our Legal Tech stack Christmas list? For us, a real estate-specific Generative AI tool with one-button solutions for running due diligence reports that has the ability to draw links between related documents. This would improve the speed and transparency of transactions (although at current product pricing levels, may not have a significant impact on direct legal costs).
In the furore of Generative AI, we think it’s worth remembering what the problem is we are trying to solve. Generative AI may not always be the answer. Digital document lists, co-authoring, and a reimagining of how we present legal reports and documents all have a part to play. And we can’t help thinking that there could be an alternative to extracting data retrospectively from legal documents we are producing.
As we move into 2025, it will be fascinating to see how Generative AI continues to evolve and shape the legal landscape. Stay tuned for more updates and insights on this exciting journey.
Come back tomorrow to open door 18 of the advent calendar and to hear from our real estate dispute resolution team on some of the most interesting cases about trespassers and protestors that have made their way through the courts this year – and the direction of travel for 2025.
Key contacts
Disclaimer
The articles published on this website, current at the dates of publication set out above, are for reference purposes only. They do not constitute legal advice and should not be relied upon as such. Specific legal advice about your specific circumstances should always be sought separately before taking any action.