Training Forward: Dr John Zerilli on reshaping law for the AI era

What excites you most about joining The Dickson Poon School of Law?

The School is in the heart of legal London, so offers unprecedented access to the courts, firms, and chambers. In addition, the wider university's commitment to AI and securing its place as a leading institution in the digital transformation now underway made it irresistible. The enthusiasm of Dan Hunter, our Executive Dean, was the final push.

Could you tell us more about your research in AI and Law?

New laws are often demanded when new technology is being discussed. Now call me old-fashioned, but anyone with an understanding of how the common law works will know very well that there’s scarcely any area of current or even conceivable developments in tech that isn’t already touched by the law. We have law in spades to deal with most of the problems we’re confronting.

Where new laws are needed, it’s often a question of effectiveness or efficiency; there’s bound to be law on the liability of manufacturers for rogue AGI, for example; but instead of relying on a set of common law precedents and existing legislation (tolerably clear as they are, in my view), it’s nice to have the matter set down in clear and unmistakable terms. And of course the other reason we may need new laws is to formalise new standards for development. But even there, the point of standards is to ensure that new technology doesn’t infringe already-existing rights. Like I say, call me old-fashioned.

What is the main problem you are solving and how do you see it fitting into the growing and evolving AI landscape?

My work has attempted to spell out how existing torts (civil wrongs) can be repurposed to deal with such mischiefs as deepfakes and disinformation online. I’ve been thinking a lot about platform governance in recent days too – both the economic justification for regulation and how it fits with the political economy of the digital platforms. I’ve been examining the notion of legitimacy, how due process (procedural fairness) confers legitimacy on government, and whether it might be a good idea for technology companies to be forced to adhere to it, even though they’re private entities (which don’t generally have to adhere to procedural fairness vis-à-vis their customers). More generally, I’ve been concerned with ensuring that the use of AI in government doesn’t breach the standard of procedural fairness we already enjoy.

Does AI have a place in legal practice?

I think it most assuredly does. It will mean, however, that the legal profession will shrink to a size that reflects need. We’ll still need advocates to argue cases in court, and some bits and pieces here and there. But I’m of the opinion that those with law degrees will become specialist kinds of humanities graduates. Their knowledge and skills will still be in demand, but not in the form traditionally packaged.

What do you think are some of the biggest legal challenges or potential legal challenges posed by AI?

A proper negotiation with the creators of digital content-those whose work is used to train LLMs. Balancing the value of the technology for the common good as against the rights of individuals to be properly remunerated for their creative labours is going to come to a crunch: either we properly remunerate those whose work feeds the LLMs or we draw a line in the sand and say, “actually, we’re faced with new circumstances, circumstances that mean if we don’t act to harness the power of technology, we may unwittingly forgo the huge promise of the latest developments.” In this sense, yes, I do believe we’re being presented with an opportunity to reflect on-and either reaffirm or reject-the current political and commercial morality that underlies the law of copyright.

Is this a field you would encourage law students to pay more attention to?

100%. I’d be having all law exams for students in my courses reflect the current and growing utility of LLMs in legal practice. Law has never been immune from technology. In the first round, about the time I was in law school some 25 years ago, word processing, electronic communications, and digitisation massively overhauled the day-to-day appearance of legal practice. I use the word “appearance” advisedly here. Suddenly, you didn’t have to visit a law library for access to case books and legislation-everything was available digitally, and all the law reports were there as faithful, searchable PDF facsimiles. But what one did in the law firm remained more or less the same, in fact, the main difference was pace and scale. Now that letters weren’t coming through the post, and everything could be sent instantaneously by email, correspondence in commercial cases took up volumes of folders. And cases that weren’t in the official reports were available for use in court literally the same day they were handed down, so a case heard yesterday was now amassing precedential value, despite it not appearing in a law report for another 2-3 years, or even ever! This round, however, is different. The nature of legal work in 2025 is changing as much as the role of the bank teller changed beyond recognition once ATMs/cashpoints became ubiquitous and no one any longer needed to go to the bank to withdraw or deposit cash. I noticed that the latest edition of 'Goode on Commercial Law' has altogether dropped the chapter on cheques, once the bane of the law student’s existence.

What excites you about the future of AI in Law?

Simplicity, ease of access, bringing the power of legal knowledge into everyone’s hands. You should only have to resort to a lawyer when you can’t come to a resolution of a dispute yourself. You should already be able to find out what your rights are without a lawyer. After all, they’re your rights! 

Tell us more about the Digital Law LLM

It’s all about "training forward". That’s the motto I’d use for the program. We’re trying to instill relevant legal knowledge in the lawyers of today with an eye to the kind of lawyers we all need to be in 10 and 20 years’ time.