The Sylvans discussed AI in the justice system based on the motion: the justice system should rely less on human judgement. Attendees debated the deep intersections of technology, bias and ethics. We explore the core arguments below.
The proposer’s opening case: using AI in the justice system
The proposer kicked off the session by clarifying the exact wording of the motion. They emphasised that relying ‘less’ on humans does not mean removing human input entirely. They noted that wholesale replacement would be madness today. Instead, they focused on the inherent flaws of human nature. People are highly fallible and carry deep biases. For example, judges often make worse decisions when they feel hungry. Furthermore, parole boards frequently display racial bias.
To fix this, the proposer suggested introducing deterministic computer systems to analyse case data. They also proposed giving judges AI co pilots. These digital tools would simply analyse complex factors and recommend specific weightings. Ultimately, adding a bit of technology will help the justice system function much better.
The opposer’s counter argument
Taking the floor next, the opposer strongly challenged the core premise. They argued that crime stems from deep inequality, not a lack of technological tools. As they stated verbatim: ‘Truly, fundamentally, we should try and create a society where there is no crime.’ Relying on AI to police society feels dangerously authoritarian to them.
Furthermore, the opposer questioned the very concept of artificial intelligence. They argued that true intelligence remains a purely human trait. When someone’s life hangs in the balance, we need full human capacity. This includes essential emotional intelligence and compassion. Since humans create computer systems, those systems must inherently carry fallibility too.
Audience contributions in the AI in the justice system debate
The debate then opened to the audience. Attendees shared a wide variety of passionate perspectives regarding AI in the justice system.
Arguments for technology and bias reduction
One speaker quickly supported the motion by comparing AI to VAR in football. They argued machines can help build better arguments as long as humans retain ultimate decision making power. Another attendee agreed with the problem of judicial bias. They cited scientific research showing how hunger negatively affects parole boards.
A different speaker viewed AI simply as a powerful tool to gather large amounts of data. This prevents activist judges from interpreting the law based solely on personal feelings. Another participant pointed out our general inability to tell when someone is lying. They suggested technology like sentiment analysis could successfully enhance the evaluation process.
Defending the human element
Many floor speakers passionately defended human emotion. One attendee referenced Dostoyevsky to argue that justice requires discretion. They stressed that justice cannot arise from rigid, deterministic rules. Automatic systems force individuals into little boxes. They reminded the room how much we all hate endless automated phone menus. Another person warned that AI relies exclusively on past data. Therefore, they described AI as ‘recycled, second rate human judgement.’
Systemic flaws and computer trust
Several attendees raised strict alarms about trusting machines blindly. One speaker pointed directly to the Post Office Horizon scandal. Massive failures happen when no one questions the computer system. Another participant stressed the extreme importance of nuance in legal settings. They shared a story of a judge recognising a defendant’s severe mental health struggles in person. A computer would miss that critical empathy entirely. Additionally, an attendee highlighted that the justice system primarily serves the ruling elite. Adding AI will just transpose existing systemic prejudices into the computer.
Democracy and the appeal process
A fascinating point emerged regarding the legal appeal process. One speaker argued that AI strongly resists paradigm shifts. Human interpretation changes over time, especially concerning vital civil rights. Removing human interpretation hinders the facility for positive societal change. Another audience member proudly defended the jury system. They felt impressed by how juries balance out personal biases. They warned against adopting AI because it relies on localised, ethnocentric data.
Closing arguments
The opposer’s final arguments against AI in the justice system
The opposer returned to frame the core issue as a theological and ethical one. They asked what kind of morality we truly want for our society. They advocated for improving human relationships rather than focusing on penal tools. To solve judicial bias, they humorously suggested feeding judges a decent Italian meal. With the right ingredients like pizza and chocolate, people naturally make better judgements.
The proposer’s closing speech
The proposer wrapped up by addressing the wide ranging debate thoughtfully. They offered a compelling historical comparison. They asked the audience to imagine a murder trial in 1850 versus one in 1950. The later trial used exciting new technologies like fingerprints and DNA. DNA evidence eventually released many innocent people from prison. Adding these scientific tools did not take away our humanity. It simply augmented human decisions. Today, machine learning can review evidence and drastically improve the overall justice system. Therefore, they urged the room to support the motion.
The final verdict on AI in the justice system
Following the robust discussion recorded in the Human judgement and the judicial system raw transcript, the Chair called for a vote. The room ultimately voted in favour of the proposition. The motion carried successfully. The idea of adopting AI and technological aids clearly resonated with the majority as a necessary step forward.
Further reading
A detailed summary and analysis of the debate can be viewed here.
Please see summaries of earlier Sylvan debates here.
For more information about how our meetings run, see meeting info.