AI: Some lawyers cannot do without it
The proportion of lawyers concerned about errors and hallucinations generated by AI has risen to 83%, the latest research from LexisNexis has found.
Meanwhile, more than three-quarters of lawyers believe AI is causing clients or the general public to misinterpret the law, while one in 10 say they are dependant on the technology.
LexisNexis’s survey found that 94% of lawyers now used AI for legal work, compared to 61% last year, 41% the year before and only 11% in 2023.
Errors and hallucinations remain the top concern. The proportion of people concerned about “inaccurate or fabricated AI-generated information” has also grown sharply, from 57% at the beginning of 2024 to 83%.
After this, the most common concern was about confidential data leakage (53%) and becoming too reliant on AI (52%).
Just over a quarter (27%) worried that ‘my quality of work will decline’, with 21% concerned about accidental bias and 9% that ‘it will increase the volume of work I need to review’.
Three-quarters (77%) of lawyers said AI was “causing clients or the general public to misinterpret the law”.
A leader at a large law firm reported a “huge increase in litigation, as claimants are able to harness AI to bring claims, not all of which will be meritorious”.
Researchers at LexisNexis gathered responses from 543 legal professionals for the report Scaling up: The AI-dependent lawyer.
Three-quarters of lawyers used AI at least once a week, with 34% turning to it on a daily basis and only 6% saying they did not use the technology at all.
Adoption was particularly high in in-house legal teams, with 89% using AI at least once a week, compared to 68% of small law firms and just over half of barristers.
The most common use of AI was for legal research (69%), followed by document summarisation (62%) and drafting and document review (both 53%).
One in 10 lawyers said they were “dependent” on AI to do their jobs. The proportion was highest in large law firms (15%) and in-house legal teams (13%).
Eight out of 10 lawyers said they felt more comfortable using AI that was “grounded in legal sources”, a “noticeable increase” from 70% a year ago.
Lawyers’ main considerations when buying legal tech were strong data privacy and security standards, transparency about how AI was used, and “human oversight of AI-enabled services”. These factors ranked ahead of cost and commercial value, or ability to explain outputs.
Less than a third of legal professionals said AI was “already embedded in their organisation’s strategy and operations”, however, with 43% saying “they are experimenting but progress is slow”.
A majority of lawyers (55%) believed AI would change how firms billed, up from 40% last year. This was particularly true for large law firms and in-house legal teams.
Dylan Brown, editor of the report at LexisNexis, commented: “The findings point to a legal profession that is becoming more comfortable with AI, but more demanding about the trust, accuracy and authority behind it.
“As adoption accelerates, lawyers increasingly want AI tools that support rather than replace professional judgement.”
Everyone will quote the 83% worried about hallucinations, but the number I’d sit with is the 53% worried about confidential data leakage — from a profession where 94% are now using AI for client work. In my experience adoption has moved much faster than the basics: knowing which tools staff actually use and what client information is going into them. A prompt containing a client’s completion details is a data incident waiting for a name. Does your firm’s AI policy say anything about client data, or just about checking citations?