AI Can Clear the Court Backlog. A Former Justice Secretary Says It Must Never Pass Judgment.

Every few months a headline promises that AI is about to clear the judges out of the courtroom. Someone who actually ran the justice system thinks that is precisely the wrong lesson to take from the technology. AI can attack the court backlog, speed up disclosure and strip the drudgery out of legal work but the moment it starts making the final call on a person’s liberty, we have given away the one thing a justice system cannot afford to outsource.

This week on the Project Flux podcast, Sir Robert Buckland KC — the former UK Justice Secretary and Lord Chancellor, now writing and advising on law, technology and the rule of law — makes the case for AI in the courts that its loudest boosters rarely bother to make: genuinely enthusiastic about the efficiency, and completely uncompromising about the judgment.

🎧 Short on time? Hit play here — it is a clear-eyed, un-hyped conversation about what AI should and should not be trusted to do in a courtroom.

Buckland’s first worry is not efficiency but evidence. Courts run on trust that what is put before them is real, and generative AI quietly corrodes exactly that. Deepfaked footage, fabricated documents and confidently invented case citations are no longer hypotheticals — they are a verification problem with a defendant’s future attached. His answer is not to ban the tools but to harden the professional standards around them: someone qualified has to be accountable for checking the machine, because a system that cannot tell a real citation from a plausible fake has stopped being a justice system at all.

Then there is bias. A model trained on decades of past decisions will faithfully reproduce the assumptions baked into them, and law is supposed to lead social change, not freeze yesterday’s norms into code and call it neutral. That is the deeper reason Buckland keeps the final judgment human: an algorithm can surface patterns and clear the routine work, but weighing fairness against precedent, and precedent against a changing society, is the part of the job that carries moral responsibility — and responsibility cannot be delegated to a system that cannot be held to account.

His prescription for regulation is just as pragmatic. Rules are written in years; models ship in weeks, and the EU, the US and the UK are already pulling in different directions. He argues for structured engagement — government, the courts and technologists building the guardrails while the concrete is still wet — rather than a verdict handed down years too late. He is candid that it may take a genuine crisis to force the issue, and that the shift to agentic AI and output-based legal pricing will arrive whether the profession is ready or not. Worth hearing before that crisis lands, not after. For the evidence base, the episode also points to a Harvard Kennedy School paper on AI, machine learning and the justice system.

Why it matters: the lesson for any regulated profession — law, audit, medicine, surveying — is the same. The competitive edge is not how fast you let AI make the decisions; it is how carefully you keep a qualified human accountable for the ones that carry real consequences.

👉 Listen to the full conversation with Sir Robert Buckland KC here.

You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.

Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.

Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.

Try Wispr Flow free

Enjoyed this? Project Flux breaks down the week’s most important AI and tech news every Monday. Subscribe to the free newsletter here and get a step ahead on the world of AI.

Project Flux is independently owned, funded, and operated by its founder. Any external references or features are for community benefit only and do not imply endorsement, control, or ownership.