Can AI Take Over the Legal System?

Impact of AI on Jobs and Industries

In 2021 and 2023 Edward Felten of Princeton University, and colleagues, modelled the impact of AI on occupations and industries. They found that AI and later large language models would have significant impacts on a range of occupations and industries. What was surprising about their research was that a number of high status industries and occupations were found on the lists.

In their earlier 2021 research one could understand the types of jobs that would be encroached upon by AI. They included genetic counsellors, financial examiners, purchasing agents, auditors, and credit analysts. All of these can be standardised and so automated. They rely on sets of repeated and routinised procedures, ideal for takeover by AI. What was unanticipated were the other jobs on the list. These included judges and magistrates, mathematicians, university history lecturers, and school psychologists. While the former list comprises innate automatability, the latter seems to be the quintessence of human-centred activity necessitating empathy, emotion, and discretion. AI may imitate such features but it doesn’t embody them. Yet there they are.

When Felten included the incursion of LLMs in his 2023 research (after ChatGPT et al entered the world) the job list changed but in an unexpected way. The list carried a large consignment of university teachers in English, philosophy, law, sociology, political science, and communications. Judges and magistrates remained present. Many, mistakenly now, thought these occupations were sacrosanct as they were focussed on knowledge production and higher reasoning.

What is important to note here is that Felten wasn’t saying that these occupations would be replaced by AI and LLMs but they would be extensively exposed to the impact of AI. He leaves this aspect open to future interpretation. We know, however, this is a unilinear path; towards greater automation.

When he looked at industries there wasn’t much difference between 2021 and 2023. Top positions were held by securities, accounting, insurance, legal services, software publishing, and artists’ agents. What distinguished the 2023 list was that legal services rose to number one as the most exposed industry to LLMs.

Why Is Law Threatened by AI

Felten’s research raises more questions than it answers. And to declare my interest here, as a professor of law and society, my main concern is with legal services and lawyers. I will use them as the example for the rest of this insight. How did they find themselves on these lists?

Lawyers will argue that AI can’t do their jobs, that law, by its nature, is too open textured and nuanced to be automated in this fashion. The subtleties of legal analysis can escape the human mind let alone the computers’ chips. Yet we know law firms and the industry as a whole is taking to AI and LLMs for legal research, drafting of contract documents, due diligence reviews, and putting together smart contracts in employment and global trade.

External observers argue that law is really only sophisticated information retrieval and data analysis, with a bit of document production (from templates) on top. Therefore, it is highly amenable to automation.  Indeed, there are current experiments with “pod” law firms which instead of having partners and junior associates, they contain senior partners and AI. AI performs the routine tasks and the partners conduct the human relationships.

Applied GenAI in Legal Practice

The 'Generative AI and Prompt Engineering for Lawyers' module offers participants a comprehensive understanding of how mastering the art of prompt engineering can yield significant benefits in various aspects of legal work. By learning to skilfully...

Learn more

The Contest between Marx and Kondratieff: The K Wave

So much of what is said here is on the whole not unsurprising: it’s easily anticipated. What is of greater interest to me is what we can glimpse on the horizon. Let me explain this using a particular framework of analysis, and to do this I need to go back to Stalin’s Russia in the 1930s.

The communist echelons hewed to the Marxist doctrine of dialectal materialism that said the relations and means of production were the crucial determinants of change in economy and society. Marx saw a sequence of societal change from feudalism through capitalism to communism. Each stage contained the seeds of its transformation from one stage to the next. We know when the Black Death (bubonic plague) struck Europe in the 1400s it wiped out half the population. Suddenly and dramatically labour became a scarce resource, and the feudal structure of master and serf came under pressure and labour were capable of demanding better conditions thereby reducing the power of feudal lords. As expanding imperialism and colonialism grew in succeeding centuries the basis for capitalist production arose. Cities and factories drew people from agriculture and the countryside to work in awful and dangerous conditions. Gradually over time better working conditions were brought in to protect children and workers from disease and accidents. Nevertheless, capitalism has proven remarkably enduring and is the universal relation of production throughout the world.

Despite the revolutions of the mid-nineteenth century it wasn’t until the 20th century that a number of societies converted to communism via revolution, in particular Russia, China, and Cuba. Their question was when would they actually transition from the intermediate socialist stage to true communism? Stalin asked a Russian economist, Nikolai Kondratieff, to come up with an explanation of how such change would occur. Kondratieff analysed agricultural prices, growing cycles and more. His final analysis led to what we know as the K wave. He argued that economies and societies went through a 60 year cycle initiated by radical innovations in technology. Previous cycles were marked by the introduction of 1. steam power (essential for early English capitalism), 2. railroads and steel, 3. combustion engines and electricity, 4. mass production and petrochemicals, and 5. Information technology and internet. The sixth wave will be characterised by artificial technology, nanotechnology, and green power. Each cycle traverses through seasons spring, summer, autumn, winter replete with growth, decline and disruption. Unfortunately, Stalin didn’t welcome this revisionist analysis of Marxism and Kondratieff spent his remaining time in the Gulag before he was executed in 1938. But his ideas live on.

Kondratieff discovered that when elites overproduced and inequality became a societal norm, disruption occurred: revolutions, pandemics, wars, and financial stringency have regularly typified the winters of the five cycles.

The New Age of AGI

After initial disparagement by conventional economists, Kondratieff’s views are being accepted, especially by people in technology who foresee the power of AI in determining the future shape of societies. I suspect that the crucial element of the sixth K wave will be a combination of quantum computing and artificial general intelligence. Both Dario Amodei of Anthropic and Demis Hassabis of Google DeepMind claim AGI is closer than we think that it will have human-like cognitive abilities in three to five years.

If this is so then the nature of work will transform dramatically. Amodei suggests that AGI will no longer just analyse data but it will create the questions, organise the research, run thousands of simultaneous trials with virtual scientists and assistants which could see the development of new medicines within months instead of years. It will do “Nobel level research”. To an extent we have seen this occur with DeepMind’s AlphaFold, the algorithm that folds proteins. What used to take the course of a PhD to solve the intricate folding of a protein has now been collapsed into a matter of moments. DeepMind has folded all proteins which is equivalent to mapping the human genome and for which Hassabis received the 2024 Nobel Prize in Chemistry.

What does this mean for legal services and the nature of law? If the prediction of two to three years for the introduction of AGI were to happen neither law nor education would have time to adapt. Unless, of course, it had operated within that context as the London School of Innovation is doing.

If we consider global law practice, for example, AGI will produce a global multiplier. Discrete jurisdictions will have little consequence since AGI will be able to incorporate them into its operations. Complex international transactions will be subject to geographic arbitrage since AGI can track regulatory changes in real time, it will be able to engage in outsourcing with great effect, and it potentially can disintermediate legal transactions enabling clients to deal directly with each other.

Alternative futures we might then want to consider are:

1. Regulatory preservation: Law societies and governments might prohibit or heavily regulate AI practice of law, creating protected professional enclaves. This seems unstable as commercial pressures and access-to-justice arguments would undermine it.
2. Human-AI hybrid: Lawyers become primarily supervisors and strategic advisors, with AGI handling execution, the “pod” law firm. This is the optimistic case but requires rapid educational transformation.
3. Bifurcated market: Routine commercial work goes entirely to AGI; only the most complex, relationship-intensive, or politically sensitive matters retain human lawyers. The profession shrinks by 80 to 90 percent.
4. Complete transformation: Legal services become fundamentally different, less about knowing law and more about: shaping business strategy, managing client relationships, providing judgment in true edge cases, or serving regulatory and oversight functions.

This is a feasible scenario for corporate law but that leaves a big question mark over access to justice. How will ordinary citizens access and use law? Will there be any human input or will we become accustomed to supercharged Amazon and eBay type online dispute resolution systems. Will private law become privatised and outsourced this way? Milan Markovic of Texas A&M law school in a recent article argues that AI, and by extension AGI, while having the potential to help those in need may also widen disparities and deepen inequality. It could give an advantage to “sophisticated actors” like landlords, debt collectors and corporations to initiate legal action. Repeat litigants will be able to pursue claims more aggressively and at higher volumes, increasing the legal burdens on those least equipped to respond. One reason for inequality here is asymmetric information insofar as people may not recognise they have legal problems rather than everyday problems. AI because of a tendency to provide confident answers may mislead people and advise them incorrectly.

Another concern with AI is how will future generations of lawyers be trained? If AI can do the work quickly and relatively inexpensively how will law firms train and charge out junior lawyers profitably? Alongside training we can ask what will happen to specialisation? Adam Smith in The Wealth of Nations (1776) first discussed how the division of labour would promote economic activity in societies by increasing productivity and economic wellbeing. The professions are exemplars of specialisation. In England we have solicitors and barristers, both lawyers but each carrying out different specific tasks and roles. The advocacy of barristers is distinct from putting together complex transactions as done by solicitors. Medicine embodies many specialisms and uses general practitioners as gateways to specialised knowledge of hearts or brains. Will these sometimes fragile distinctions be eroded by AI? It is already happening in some areas. Accounting has adopted combinations of data collection on blockchains which are in turn analysed by virtual AI auditors. These kinds of decisions then blend into the consulting side of accounting firms and feed their activities.

If AI, or more specifically AGI, can have an all-seeing eye over fields of activity and delve deeply into particular aspects, what exactly would be the role of the narrow specialist? Would it not be possible for AI to encompass many areas of knowledge and cross-fertilise between them potentially coming up with new insights and methods? They are questions we will have to face as the sixth K wave rushes upon us.

I conclude stating that I don’t see this is as either a utopian or dystopian vision. As we enter the sixth K wave disruption will occur. People will need to be educated and trained to adapt and learn how to innovate in this new era. If we can befriend AGI and not be threatened by it, the new era could be productive, fruitful, and beneficial. I hope so.

MSc AI for Business Transformation

Artificial intelligence is revolutionising the way businesses operate, creating new opportunities and challenges; This programme is designed to equip you with the expertise to navigate this evolving landscape with confidence and precision. You will...

Learn more