Many believe AI will soon take over most legal tasks, but Mark Doble finds that optimism misplaced
Public trust in AI’s legal decision-making capabilities remains low, with only one-tenth of Canadians comfortable with AI making legal decisions, according to a recent survey.
This skepticism is not only justified but necessary, says Mark Doble, CEO of Alexi, a Canadian legal tech company that conducted the survey along with the market research company Talker Research.
“We strongly believe that AI should not be making key legal decisions. It can influence the decisions. It can participate. It can help humans make these decisions, but the legal system and rights and liabilities are fundamentally human, and that means humans need to be the overseers of those outcomes. It’s really good and promising that only 12 percent think AI should take over legal decisions.” he says.
Doble says AI should be designed to support legal professionals, not replace them.
“Our legal industry is designed for humans… Rights and liabilities and obligations – these are very human things. We need humans to be the ultimate bearers of how these things get enforced.”
The survey, which polled 2,000 respondents aged 18 and older in the US and Canada, found that over one-third believe AI will take over most legal tasks by the end of the year.
Their confidence, Doble says, is unfounded.
“In the industry, there’s a much more sophisticated view on the role of human lawyers and the role of AI,” he says.
He believes that lawyers will have a clear role in the future and that AI will never fully replace them.
“But we also believe there’s probably a much bigger role for AI than what most people in the [legal] industry actually believe,” he adds.
He says AI will transform lawyers' work rather than replace them, taking over repetitive, verifiable tasks while leaving strategic decision-making to human professionals.
“If an AI tells you something, you should be able to go and verify: is that right or not right? Anything that’s verifiable by a human professional, we should have AI actively involved in supporting those roles.”
Legal research, document review, contract analysis, and case law summarization can be automated, but core legal functions – like advocacy, advising clients and using social and emotional intelligence to present your case – will always require human judgment, he says.
While AI can’t be expected to replace lawyers fully, and it should not be trusted to make legal decisions, it could soon become a trusted mediator in legal disputes, Doble says.
“We’re very clearly on a path where binding arbitration decisions will be mediated entirely by AI. This is a very clear possible outcome,” he says.
He says AI-driven arbitration could streamline dispute resolution, making legal outcomes faster, more affordable, and objectively fair. However, he adds that human oversight will still play a crucial role even in AI-driven arbitration, especially for appeals.
“Ultimately, if you still disagree, there is a path to appeal… And I have no doubt that AI arbitration will be a very prominent fixture in society in the very near future.”
Beyond arbitration, he predicts AI will become more integrated into legal workflows and handle much of lawyers’ daily communication and administrative tasks.
“AI is going to obfuscate products like email. Pretty soon, lawyers probably won’t even be the ones sending emails… They’ll just be interacting with AI all day, every day, and the AI will be doing the rest.”
Doble says the two most significant barriers to a broader implementation of AI are misconceptions about its reliability and the rapid pace of technological change.
Many legal professionals are wary due to past issues, such as AI-generated hallucinated case law, where non-existent cases were mistakenly cited in legal filings. However, Doble says significant progress has been made in eliminating these errors.
“We’ve learned so much in a year. Hallucinated case names were a problem before, but we’ve now developed techniques to significantly mitigate or even eliminate them.”
Another key challenge, he adds, is keeping up with AI’s rapid evolution.
Even some of the world's largest law firms are often surprised by what is now possible, Doble says, as many advancements available for months – sometimes even a year – are only just coming onto their radar.
Contrary to the common belief that lawyers are slow to adopt technology, Doble says that the real issue lies with subpar legal tech products that have historically failed to meet lawyers’ needs.
“People are quick to blame lawyers for being resistant to technology, but that’s not true at all,” he says. “We've got a couple of decades now of legal technology that has built … products that were hard to use, hard to trust, and hard to rely on. When lawyers see a tool that actually helps their business, they adopt it.”
