Artificial intelligence (AI) tools are rapidly transforming legal workflows – from accelerating research to streamlining contract analysis and enhancing e-discovery.
But, while these tools offer speed and pattern recognition, legal professionals caution that they cannot replace the strategic judgement, foresight and nuance that define quality legal practice.
Safee-Naaz Siddiqi, senior associate in Cliffe Dekker Hofmeyr’s (CDH) Knowledge Management and Competition Law practices, references a study by Vals.
“The Vals Legal AI Report (VLAIR), which benchmarks legal AI tools like Harvey and CoCounsel against human lawyers on real-world tasks, shows that AI tools consistently outperform lawyers on tasks like document Q&A, summarisation, and transcript analysis. But the study showed that humans still had the edge when it came to redlining and deeper legal research.”
Moreover, the risks associated with AI use requires human oversight in all tasks, even those that the AI tools can ostensibly deliver more efficiently.
In this sense, Siddiqi says AI is forcing lawyers to rethink how they tackle workflows. “At CDH, and especially in Knowledge Management, we’re asking questions like: what should lawyers be spending time on? Where can we shift junior training now that certain kinds of work are able to be automated? And how do we make sure lawyers are being properly trained to use AI responsibly?”
Siddiqi describes how AI tools have already significantly improved efficiency across time-intensive tasks. “In the past, practitioners would spend hours wading through databases to locate relevant case law. Now, AI-powered legal research platforms allow lawyers to ask questions in plain English and receive targeted results more quickly than traditional methods.
However, these tools aren’t perfect – the results still require careful review and verification by legal professionals to ensure accuracy and relevance. When used properly as a starting point for research, they can significantly improve efficiency in the initial stages of legal research.
“So, it’s not just faster – used correctly, it’s sharper,” she explains. “The tool does the initial sweep, so lawyers get to spend more time thinking about the results. The three, four, sometimes ten hours you would spend trying to find the cases, is now spent on analysis and strategy.”
Similar benefits have emerged in litigation, where AI is helping surface the most relevant threads in eDiscovery, and even map sentiment and patterns that would be hard to detect manually. But Siddiqi is clear that these tools do not eliminate the need for human legal professionals. “If anything, they emphasise where human insight is indispensable,” she says.
Siddiqi goes on to caution against the misconception that adopting legal tech means lawyers need to become coders or data scientists. That said, it is also not good enough to apply the technology without a sufficient level of understanding. “It’s not about knowing which button to press. It’s about judgment – the ability to understand what the tool that you are using is optimised for, where it falls short, and how to make informed decisions around the use of these tools.”
One of the most pressing shifts, she argues, is in how lawyers are valued – not for the volume of output, but for the quality of strategic thinking. “If all of us now have access to the same AI tool, and it takes us five minutes to prep a discovery affidavit, what will separate the good lawyer from the average lawyer?” she asks. “It’s the one who spends more time on the substantive task – thinking 10 steps, 20 steps, 30 steps ahead. That’s where legal professionals offer the most value, because AI thinks more literally than laterally.”
Siddiqi adds that this is particularly true in high-stake matters. Drawing on her litigation experience, she explains: “When I get a letter of demand from someone, I’m not just thinking about responding; I am thinking about how my response would look if I had to present it up in the Supreme Court of Appeal.”
While AI tools may produce technically correct answers, they lack strategic thinking about leverage and positioning. “They can’t assess what really matters to your opponent and adjust strategy accordingly,” she notes. “They’re not thinking about these broader tactical considerations.”
This is why CDH’s training strategy is focused not just on how to use AI tools, but on how to challenge and interrogate their outputs. Lawyers must be able to look at an AI-generated clause and say, “This is why the difference is better”, or reject it where appropriate. Data literacy, too, is becoming essential, not in the sense of coding, but in understanding what drives an AI’s recommendations, how confident to be in its output, and when to override it.
“Ultimately, AI is a tool,” says Siddiqi. “Holding a katana won’t make me a samurai. What separates the great lawyer from the average one is still their ability to think, to strategise, and to see what’s coming next,” she concludes.