“Modernising Criminal Justice” 2025: the Coming of AI

by Mike Nellis, Emeritus Professor of Criminal and Community justice at the University of Strathclyde and Clerk of Quakers in Criminal Justice.

This blog post reflects the author’s opinions and is provided to inform and stimulate debate;
it does not necessarily reflect the position of JAAG.

I have been irregularly attending the annual one-day Modernising Justice conferences in London (and their Digital Justice equivalents in either Edinburgh or Glasgow) for nearly two decades, but this year was the first time I’ve been invited to speak at one. Way back, my interest was in the ethics and politics of electronic monitoring (EM/tagging) and its utility or otherwise for the probation service, but these conferences got me interested in technology and criminal justice more broadly, whether virtual courts, predictive policing, the use of live facial recognition and now artificial intelligence. Though I wear my “emeritus professor” hat at these events, not my Quaker one, I have been trying throughout to work out what a Quaker perspective on these developments might look like.

The conferences are organised semi-independently of government but aim to showcase and promote government policy. All are premised on the idea that new technologies are indispensable to modernising decrepit criminal justice agencies. Some 500 senior and middle managers from all agencies and some tech companies attend them.  The conferences take place in the vast QE11 Conference Centre in Westminster, which has ample room for all the national and international vendors who sponsor them, some well known - IBM and Amazon – but most not. These either deliver advanced digital services to government agencies already, or hope to. Their product displays, affable salespeople and glossy brochures (or, nowadays invitations to scan QR codes) illuminate the upcoming brave new worlds which they are keen to build and profit from.

                The conference theme on 12th June 2025 was the use of Artificial Intelligence (AI) in criminal justice. Ever since we began wondering, years ago, what the unspoken endpoint of incremental digitalisation in governance processes might look like, AI has hovered on the edges of our expectations, for better or worse. AI, loosely used to denote technologies which automate cognitive processes hitherto undertaken by people, has been mooted at previous conferences (in Scotland too) without it ever being clear that it would eventually be significant. This year in England, in the slipstream of a full government commitment to AI as a driver of both economic growth and the transformation of public services, AI was front and centre.  I am now ruefully convinced that AI will indeed be infused into criminal justice processes, initially to improve efficiency – and who can be against that? -  but I have grave doubts about what it will do to the humanity of the services in question. Furthermore, I expect it to exacerbate rather than remedy the inequalities with which criminal justice has long been riven, and to leave untouched the social and psychological causes of criminality.

                The Modernising Justice conferences always take some account of whatever criminal justice crisis is current at the time, and mould their putative tech solutions to them. This year was no exception. Former Lord Chancellor David Gauke was interviewed about his recently published Sentencing Review in England, and Lord Timpson, the minister for prisons, parole and probation, delivered a powerful keynote speech. Gauke’s Review is less the “landmark” that even The Guardian foolishly called it, and more the last best hope for a broken system, albeit only a slender one, if its all too modest aim -  a temporarily reduced rate of prison use by sentencers - is to be achieved. To his credit, Gauke opened by saying that he had never seen any good reason why the England and Wales prison population should have grown from 40,000 to nearly 90,000 in the last four decades, why maximum sentence lengths had been increased so dramatically or why Britain should have topped the West European league tables for prison population size for quite so long. He was not asked to explore how or why this dismal state of affairs prevails, or whether – despite optimistic claims to the contrary - his own proposals are actually more aligned with them than he might care to admit.

Gauke conceded that his Review had been commissioned not out of principle but out of necessity to forestall another embarrassing emergency release programme of the kind that the new Labour government had been forced into last autumn.  This is mostly to be done by encouraging sentencers to use more punitive versions of electronic monitoring instead of prison sentencers, and as part of post-release programmes, which may then enable earlier release, thereby reducing prisoner numbers. While he was adamant that rebuilding the probation service to its former levels of professionalism was vital, and paid suitable lip service to its rehabilitative capabilities (despite this not having been in the initial remit of his Review) we have every reason to be wary of the kind of probation service that might actually emerge.

                Lord Timpson also made the funding and strengthening of the probation service central to his keynote, while making explicit that whatever comes from Gauke’s Review is essentially a stopgap, pending the delivery of 14,000 more prison places by 2030. Like Gauke, Timpson is keen on technologically augmented “community sentences”, which are no longer being called alternatives to custody, or even community penalties, but “prisons outside prisons”. None of this is hopeful; quite the opposite. Apart from the unlikelihood of sentencers making use of them on sufficient scale to slow down the rate of increase in prisoner numbers, these grim new measures are likely to remain in place even after the nirvana of 14,000 more prison places has been achieved. This will transform the landscape of community sentences and probation for the worse: their future forms may have only limited resemblance to the rehabilitation-and-risk focussed agency it once predominantly was. Rumour has it that senior people in the Ministry of Justice believe  the recruitment and retention crisis in probation can be solved by making it a more digitally-infused service, whose new controlling capabilities will inspire and attract a younger, tech-savvy  generation of officers.   

                Government spokespeople are routinely confused when they speak glibly of combining punishment and rehabilitation, as if these were invariably compatible and it was self-evident when and how the balance should be struck.   Unduly onerous punitive regimes tend to alienate people who offend, and disincentivise them from engaging with such rehabilitation programmes as they are then offered. Lord Timpson himself seemingly fails to grasp this, and now entertains imposing levels of coercion on some offenders in the community akin to those prevailing in prison, with all the risks of increased recall that this entails.  

The danger of punitiveness was the focus of both the verbal submissions I had been asked to make to Gauke’s committee. The first was an ethical-polemical plea against developing more punitive forms of EM in order to create “prisons outside prisons”, potentially by increasing the hours of monitoring,  lengthening the time served on the order, and adding more behavioural prohibitions. The second – to civil servants not the whole committee  - was a detailed exposition of the international evidence base on EM, which is clear that the best use of it – in terms of reducing reoffending – is primarily as a form of control alongside rehabilitative programmes. Alcohol monitoring – which is currently used to ban binge drinkers who offend from drinking for several months is the one punitive exception which seems to work, at least in terms of compliance with what courts require. An equivalent form of monitoring for drug consumption may soon be available, but these are not, I think, precedents for other forms of EM: they demand very specific, easily checked forms of compliance from offenders rather than acquiescence with a wider regime of the kind implied by “prisons outside prisons”.  It turns out, however – from another post-Gauke interview with civil servants to which I was invited - that fleshing out the concept of “prison outside prison” remains a work in progress in the Ministry of Justice, which may include using remote-controlled “smart home” technology to impose greater intensities of domestic confinement.

                Apart from the high-profile individual speakers at the Modernising Justice conference – Sir Brian Leveson also spoke sagely about the unprecedented backlog of cases in the English Criminal Courts and urged conference attenders to find some “technological” means of reducing it - most of the day consisted of plenary panels of speakers and a selection  of workshops. These were mostly made up of senior civil servants and experts from tech companies. The panel and workshop titles variously spoke of “leveraging”, “harnessing” or “optimising” AI in various settings, the necessity of gathering and processing more data at all levels of the system, applying “decision intelligence”, integrating innovation, AI in prison education, and learning lessons from building a chatbot to advise litigants in the Family Court. Those I attended were universally illuminating and if I was uncomfortable with the unspoken assumptions behind the new approaches, I came to understand the government’s direction of travel better than I otherwise would have done. 

I was one of four people on a plenary panel concerned with “navigating the responsible and proportional use of AI in criminal justice”. I seem to have been invited because I was known, from other things I have recently said and done, to be a sceptic about developments in AI, and that I would speak against the designed-in, upbeat tone of the event as a whole. I duly did this, questioning especially the benefits that AI would allegedly bring to our under-resourced and overstretched probation service. At the moment generative AI (in the form of recording, transcription and translation services) is being piloted in the expectation that it will save probation officers’ time, which they might hopefully spend on the more important, “professional” tasks with their clients. Even the Council of Europe’s recent (October 2024) ethical recommendation on the uses of AI in prisons and probation, which is extremely positive and optimistic about AI, doubts that this will last for long, considering that it may instead be the precursor of automating probation’s  professional tasks as well (which, though feasible, it does not favour). As it happened, one of my fellow panellists was a professor from The Alan Turing Institute who has recently authored an ethical policy statement on AI specifically for the Ministry of Justice, and we agreed on a lot of the questions that need to be asked of AI before we got to considering what is “responsible”  and “proportional”, though we did some of that too.

Quakers have a long and honourable tradition of working in criminal justice and penal reform, grounded in spiritual and humanistic values. Although AI will increasingly affect the process and practice – and probably the ethos - of familiar agencies, much will remain the same in terms of victims and offenders’ human needs: it is the system’s needs to which AI is catering, however it is dressed up for public consumption.  AI, we must remember,  will be rolled out alongside the continuing, morally indefensible expansion of the prison estate and the increased punitiveness of community penalties: in some instances AI will actually be complicit in this. Even if AI has some practical benefits – for example in in-cell prison education – tiny gains, however dazzling, will not redeem the vast enduring failings of the British penal system. We must not assume that “modernisation” always means improvement, or debate the future as if AI was the only thing shaping it.

 Quakers and other faith groups certainly need to engage with these new developments but in a clear-eyed way, recognising that the emerging penal agenda is as much influenced by the aspirations of tech companies as by consideration of the common good  - or what love and justice requires. It was helpful that Phil Bowen, director of the Centre for Justice Innovation in London gave the keynote speech at the 2025 Quakers in Criminal Justice conference, to orient us towards these challenges. An earlier meeting between JAAG and QICJ members had paved the way for this. It is also encouraging that in the USA the Quaker Institute for the Future has begun addressing AI. But these are the just first steps in a long journey.