In-Home Counsel
How a lot do authorized leaders belief synthetic intelligence in high-stakes choices? New research sheds gentle
Solely 37% of authorized leaders belief using generative synthetic intelligence in high-stakes choices, exhibiting restricted confidence in its capacity to interpret complicated points, in line with a brand new research of 500 authorized and enterprise leaders. (Picture from Shutterstock)
Solely 37% of authorized leaders belief using generative synthetic intelligence in high-stakes choices, exhibiting restricted confidence in its capacity to interpret complicated points, in line with a brand new research of 500 authorized and enterprise leaders.
The study by Paragon Legal, a authorized providers firm that advises companies and company authorized departments, additionally reveals that:
• 39% say their organizations are adopting AI too rapidly.
• 36% have used AI-generated insights that they don’t absolutely belief.
• 37% have restricted or disabled AI instruments due to considerations over compliance.
“As synthetic intelligence turns into extra deeply built-in into authorized observe, leaders are more and more navigating the place to attract the road between innovation and integrity,” in line with the research. “Belief in AI stays cautious, particularly when human judgment and accountability are at stake.”
Practically two-thirds (65%) of authorized and enterprise leaders who responded to the survey say AI needs to be used for authorized help, moderately than attorney-led choices. Nevertheless, the research additionally reveals that 41% admit that they’ve reclassified duties to help with AI adoption.
A key takeaway for authorized leaders is to “set up clear accountability frameworks earlier than permitting AI to affect substantive authorized choices,” in line with the research. It additionally means that “human sign-off ought to stay the usual, particularly the place regulatory interpretation or consumer influence is concerned.”
Write a letter to the editor, share a story tip or update, or report an error.