At the risk of spiralling off on a personal ramble, I wanted to set the scene by recommending a podcast “Sleepwalkers”, an iHeart Radio production hosted by Oz Woloshyn and Karah Preiss, and the book Thinking Fast and Slow¸ by Daniel Kahneman.

“Sleepwalkers” is an exposition on the latest developments in artificial intelligence (AI). The title implies the gist of the show. Developments in AI are much more advanced than most people realise. Right now, most of us are “sleepwalking” into the future, more or less oblivious to the fact that we already have deep learning “neuro nets” that can compose music, write novels, and will likely pass at least aspects of the Turing Test. Since listening to the whole series, I have learned that Estonia is already trialling AI to resolve small claims disputes. The prospect of an AI system as judge is, at the very least, remarkable.

Frankly, I am now firmly of the view that those who ask “will robots take our jobs?” risk asking the wrong question. The relationship between human beings and productive labour per se is on the cusp of total disruption. In the immediate term it may be as simple as some AI replacing finite tasks and/or roles. However, the changes that loom are much deeper than merely replacing accountants with an AI bot who knows about finances. The world will not need accountancy, if AI does away the financial system … but that really is reaching off into the future, and well beyond me and beyond LawTalk.

Moving to the book, Thinking, Fast and Slow is the most accessible examination of human psychology and judgement I have read. Kahneman is a professor of psychology and Nobel Laureate in Economics. Thinking, Fast and Slow was a best seller shortly after publication in 2011. The book systematically (and I say convincingly) shows the demonstrable shortcomings in our instinctive judgements. Common fallacies and biases are laid bare.

For example, Kahneman relays a study that showed parole decisions (made judge-alone) tracked consistently with the judge’s meal breaks. Even accounting for all other variables, and the days prior that judges had to review the files, approvals peaked after each meal break, and declined as the time since the judges' last break increased. This sort of information does not negate the value of human judgement, but it does cast it in a very different light from what most of us assume about how judgements are made.

You put the podcast and the book together, stir in with over a decade of prosecuting jury trials, and you get this: not so much an article as a suggestion that practitioners think about how cases are decided.

My questions

This topic is incredibly broad, but as a criminal practitioner I naturally lean into it around that most ancient of institutions, the jury trial. Unhelpfully, I am not about to advocate any sweeping change nor ‘the answer’. So please stop here if you thought you were about to be enlightened. All I can do is set out the questions that I have, in the hope that these might spark your interest.

First, I need to be clear that I am not about to attack the jury trial itself. Some level of community engagement is essential for meaningful justice. I am always disappointed when the response to a jury’s supposed weaknesses is to advocate for judge-alone trials. Judges are no more nor less human than jurors. Experience of the law is both a gift and a curse when it comes to determining questions of character and fact. To presume that one judge (and every such lone judge) could magically overcome a bias we fear in 12 lay people is a bridge too far.

Second, in the confines of this brief article, I cannot advocate for anything more than that classic soft option, “please think about how criminal cases are determined”. When/if opportunities arise, I hope that lawyers (and not just litigators) will engage to try and improve our justice system. Without wanting to be too melodramatic, I genuinely fear that if we do not get more engaged in how criminal cases are determined, then the prognosis for the criminal courts is poor. If we reach a point where the wider community lose all faith in our criminal justice system, this will have dire consequences.

Turning to the questions I wanted to pose (in no particular order):

1. Why is resistance to jury service endemic?

When people find out I am a criminal lawyer, the two questions they inevitably ask are variations on, “how do I get out of jury service?” and “how can you/your colleagues defend people they know are guilty?”

The latter is a topic for another day. The former speaks to a widespread and heartfelt resistance to serving on a jury.

This is in part financial (below), but money does not entirely explain the question. People that I quiz about their reluctance cite emotional strain, reluctance to sit in judgment on others, and fear of intimidation. At least some of these are also reported in juror research in New Zealand (below).

The seemingly natural tendency to want “someone else” to determine cases is understandable. For most lay people, this means expecting the judge to guide them to the “right” answer. It is this that leads many to shy away from proposals around shared judge and jury deliberations. Many understandably expect that the jurors would be easily over-influenced by the judge.

It is not a quantum leap from wanting “someone else” to do it, and the pressures on our bench and justice system, to moving some aspects of judgment to AI.

For those who think I am dreaming, we already see this occurring in limited ways. I understand that some US states rely on an algorithmic assessment to assess risk and therefore determine parole eligibility. As above, Estonia is reportedly trialling an AI system to determine small claims. I genuinely fear for our civil society if our impulse to want “someone else” to decide cases evolves in a way that undermines or even removes the role of human judgement in trials.

Can we address this tendency to want “someone else” to do the judging? It is safe to assume we cannot get rid of the impulse, but are we doing enough to build confidence in jurors and equip them for criminal trials?

2. Should we start identifying the obvious risks and adjusting our system to mitigate them?

It is difficult to “know” anything conclusively about human psychology. Still, surely we now know enough to act in response to how deeply flawed human judgement can be.

Shooting from the hip, we are better equipped than ever to realise risks including:

• Conscious bias. Even seemingly random selection from the electorate does not generate truly representative juries. Combine this with the capricious “challenge” system during balloting, and it is difficult to assert that the jury is somehow genuinely a random cross section (and therefore free of, or at least balanced, in terms of prejudice).
• Unconscious bias. We are consistently finding out an alarming array of subconscious influences on our judgements.
• Flawed/misplaced trust in intuition. This is at its most stark with assessments of character and credibility, and these often go to the heart of a criminal trial.
• Flawed, and in fact deeply inaccurate, lay “understanding” of human memory, conscious experience, and “normal” reactions to stressful situations.
• An inability to retain and process information presented orally.
• Related to the above, dramatically shorter genuine attention spans than we perceive that we have.

Given these and other risks, it is remarkable that we still:

• Hear trials in roughly four long bursts each day, averaging over an hour per session.
• Rely predominantly on oral evidence, albeit supplemented with pictures and a transcript (the latter often provided much later than the oral evidence).
• Prioritise evidence given at the trial itself, in preference to accounts from closer to relevant events.
• Permit submissions that play upon well-known falsehoods, for example, the well-rehearsed “rape myths”, as well as known (but nonetheless popular) fallacies about human memory like, “if X really happened, you would expect the witness would remember what the offender was wearing …”

Some may advocate AI to supplement or even replace some aspects of flawed human reasoning. This is dangerous territory. First, there is “automation bias” – we assume that what the “machine” tells us is right, because it is an “objective” assessment. The problem is that whoever/whatever programmed the machine can build in biases every bit as pervasive as classic human foibles. Second, the ethical ramifications of delegating important decisions to a non-human agent are huge.

3. There appears at least some limited consensus for tempering the jury trial process for cases involving allegations of sexual violence. Should we really limit reform to cases of sexual violence?

The work in this area is compelling, and instinctively many people seem to agree that sexual violence forms a special class of allegation, requiring special sensitivity.

Pragmatically (ie, practical impact on complainants, defendants and witnesses), I agree that extra sensitivity is required. However, at some point the principled difference between a sexual allegation and any other serious criminal charge vanishes.

If incremental change is the order of the day, then so be it, and prioritising the most sensitive cases makes sense. Still, if we find ourselves agreeing that we can do better than jury trials for sexual cases, and/or that we can improve how we do jury trials in these cases, then I suggest we must consider whether those same improvements should be rolled out to all criminal trials.

4. Can we really say that it is enough to pay juror’s parking/travel expenses and a token $62 per day?

It is safe to assume that paying jurors more is fraught, particularly if every trial requires 12 jurors.

Do we “need” 12 jurors? Can we use less, and spend more on making them more likely to be effective judges of fact? For example, would it be better to use less jurors, but accommodate or transport them for the duration of the trial (and pay them nothing directly)?

Is it time to find some middle ground, between the legal experts (judges) and lay people (jurors)? Again, just for an example, should every citizen receive some training before sitting as a juror?

Before I go…

Before signing off, I wanted to highlight the Law Commission’s earlier projects in this area:

“Juries”, commenced in 1999 now closed, which spawned the 2001 Report 69, Juries in Criminal Trials.

“Alternative Models for Prosecuting and Trying Criminal Cases”, commenced in 2009 now closed, which resulted in the 2015 Report 136, The Justice Response to Victims of Sexual Violence: Criminal Trials and Alternative Processes (JRVSV). 

If I have piqued your interest, then I encourage you to get your hands on these thoroughly researched and well-presented reports.

I was also pleased to see recent academic work available online (albeit focused on the area of sexual violence, when I wonder if it’s time to re-ignite earlier reviews of the trial process regardless of the “type” of crime):

Camille Wrightson “Judging Juries: Assessing a New Fact-Finder Model for Sexual Violence Trials” Bachelor of Laws (Honours) Dissertation, University of Otago, 2017.

Amelia Erin Retter “Thinking Outside the (Witness) Box: Integrating Experts into Juries to Minimise the Effect of Rape Myths in Sexual Violence Cases” (2018) 49 VUWLR 157.

Given what I am talking about is so fundamental, and that “good things take time”, I am not exactly making a call to arms. However, I would encourage all practitioners to consider our criminal justice system critically.

This is not exactly new territory. However, there are fundamental changes ahead, including the rise of AI. Even putting aside AI, we are simply not the society we were when the current jury system was conceived. With more demands on our time and attention than ever before, it is high time we at least check and make sure that our criminal justice system is not merely “sleepwalking” along.

Chris Macklin chris.macklin@gordonpilditch.co.nz is a senior Crown Prosecutor with Gordon & Pilditch. He is also a member of the New Zealand Law Society’s Criminal Law Committee. The views expressed in this article are those of Chris.