Unlike generative AI, which creates content, agentic or autonomous AI can take action on behalf of users, interpret goals, plan and adapt their own actions, and execute multi-step tasks independently.
It is not currently used in the justice system ‘but looks likely to emerge’ gradually, driven by commercial pressure, according to Law Society research published this week, ‘The future of agentic AI in the justice system’. This risks systems capable of acting on our behalf becoming enmeshed before regulatory controls to preserve human judgement and accountability are in place.
Currently, for example, judges can use AI to support their work but remain the decision-maker and accountable. The research paper notes that agentic AI reorganises how the decision is formed because it sets the framework by assembling materials, identifying authorities and structuring lines of reasoning. The research paper warns: ‘The risk is that it becomes more difficult to understand where judicial judgement ends and an AI system-generated analysis begins.’
Another difficulty highlighted in the paper, if assigning routine work to AI, is how to decided what counts as ‘routine’—making this decision requires judgement about context, risk, vulnerability, discretion and potential consequences. ‘However, a case can look routine because relevant context is missing, or the system has filtered out some complexities. This may result in delegation to AI growing subtly, instead of through an explicit transfer of judicial authority from humans to autonomous systems.’
Law Society vice president Brett Dixon said: ‘Legal judgement cannot be delegated. Not in courts and not in the legal sector. There is still time to shape the boundaries of AI before the market does it for us.’
