The future use of artificial intelligence in courts should be shaped by public consent, legislation and careful attention to how people experience justice, rather than by predictions of inevitable disaster. AI could make legal services faster and more affordable, but human judges are unlikely to disappear, particularly in cases involving families, children and complex commercial disputes.
People who submit perjury to a court will be found out and sent to prison, just as they would be if they had forged a will. At the same time, predictions about technological disasters should be treated cautiously. “I think we need to be careful when predicting disasters. I’m very optimistic,” the discussion says.
Building a justice system people will accept
The prospect of simpler civil cases being decided entirely by AI has been described as an unavoidable future. But the assumption that courts are inevitably moving in that direction deserves scrutiny.
Legal systems need to find ways to deliver justice that fit the culture in which they operate. That does not mean that all litigation should be conducted on TikTok. However, people increasingly turn to large language models for answers to almost every kind of question, including legal questions. Ignoring that reality could lead to mistakes when designing a process suited to the 21st century.
Acceptance of algorithmic decision-making will depend on consensus. People can be asked to accept an algorithmic process in particular cases only if they consent to it, or if society establishes its use through law.
When people receive an unfavorable result in court, they may eventually accept it—even if they believe the decision was unusual or wrong—if the process has been conducted humanely. AI-based decisions may be harder to accept when people feel that the outcome has been imposed by a machine.
“I would really like to see a very open international discussion about how conflicts should be resolved,” the discussion says.
The challenge of explaining AI decisions
One reason people oppose AI trials is uncertainty about how a model reached its decision. The explainability problem is not necessarily a question of logic, but of outcomes.
AI can produce highly plausible inferences even when its internal reasoning is difficult to follow. Although it may not be possible to see how a model generated a particular inference, people can still examine the inference itself and judge whether it is valid.
Human judges are unlikely to be abolished. There is something almost biblical about human judges, and human psychology encourages people to want to be judged by their peers. Nonetheless, in some situations people may prefer immediate results, even when those results do not come from a human decision-maker.
Where should the line be drawn?
The judiciary is not yet in a position to determine the limits of AI in legal decision-making. That is a question for society, politicians and the law, making an open public discussion especially important.
Child-custody and other family cases are among the areas that essentially require human judgment, including in the view of younger generations. The same may apply to complex commercial cases. AI could nevertheless help gather and organize information, making the decision-making process more efficient and easier to manage. That could significantly change how commercial litigation is conducted.
Used in that supporting role, technology could give more people faster and more affordable access to justice—an outcome that would benefit both the economy and the people the legal system serves.
Source: www.wired.com
