The challenge is not just to go faster, but to go together – Karyn McCluskey

Community Justice Scotland’s chief executive Karyn McCluskey discusses how AI tools can be useful – but could cause problems in the justice system if not properly managed.

The world feels as if it is accelerating beyond our collective stride. What were once described as technological ‘leaps’ now look more like chasms, sudden openings that leave institutions, workforces and communities scrambling to keep pace. It’s not simply that innovation is moving quickly, it’s that our ability to anticipate its consequences has not kept up.

That matters acutely in justice. Horizon scanning is always difficult, but it is harder still when one part of the system gets excited about a shiny new tool, a promise of speed, or that automation can remove work from stretched services.

Recent media debate around the Metropolitan Police’s proposal for Palantir technology is a case in point. The promise sounds compelling: artificial intelligence that automates elements of intelligence analysis, helps investigations move quickly, reduces bureaucracy, detects more crime and improves outcomes for victims and witnesses. It is hard to argue against. Many will think it sounds like a tip top idea.

And it is only one part of a wider technological shift in justice. Live facial recognition, for example, can be used to prevent crime, identify suspects and protect vulnerable people. In real time, it can flag people on watchlists, assist officers when someone gives false details, or help identify someone who cannot communicate. After an incident, it can match images against databases, speeding up work that once took far longer.

It sits within, and acts upon, a complex justice system. When one part of that system becomes more efficient, the effects ripple outward.

If AI and facial recognition enable more crimes to be detected and processed, more cases will flow into the courts. More investigations can mean more arrests. More arrests can mean more prosecutions, more sentences and, often, more demand for custody or community-based supervision. In the short to medium term, the likely trajectory is clear: more throughput. There may be an assumption that deterrence will offset this, but that is neither immediate nor guaranteed.

That creates a fundamental challenge. While parts of policing may be enhanced by automation, the justice system remains profoundly human at its core. Courts rely on judges, lawyers and administrative staff who are already stretched. Community justice services manage rising caseloads with finite resources. Prisons are operating beyond capacity. These are not elastic systems that can simply absorb a surge.

The risk is that we create a bottleneck. By accelerating the front end of the system without strengthening what follows, we open a sluice gate upstream and leave downstream services to manage the flood. The result can be delays, poorer decision making and, paradoxically, a worse experience for victims and communities.

None of this is an argument against technological progress. It is an argument for thinking system-wide before we press go.

Investment, workforce planning and system design need to move in step with technological change.

The challenge is not just to go faster, but to go together. Without that, even the most sophisticated tools risk overwhelming the systems they are meant to improve.