The End of a Legal Ice Age

Remember when ice had to be delivered from afar, not effortlessly produced in your own home?

Me either.

I do recall an insulated box on my family’s front porch, into which cold bottles were regularly deposited by the ‘milk man.’ And I often walk around a pond that was used for ice harvesting about 150 years ago.

For most of human history ice was a luxury, requiring serious labor and transport. (See e.g. Ice! The Amazing History.) But the iceman no longer cometh. (Eugene O’Neill’s 1939 play, The Iceman Cometh, tells the story of alcoholics who live in a flop house above a saloon.)

I ended a 1990 talk with this:

Law is going to look a lot different in the next century — to both public and practitioner. Trying to say exactly how is like blowing smoke rings in a wind tunnel. … But there’s one thing I’m reasonably sure of: in the not-too-distant future more than a few lawyers are going to be echoing the sentiments of John Astor, who, seated at the bar on the Titanic, supposedly quipped “I know I asked for ice, but this is ridiculous.”
(From The Computer as Lawyer-Saving Device)

Much of what we lawyers do can now be done more conveniently and inexpensively by knowledgeable machines. The media are overflowing with examples. Sure, there are many ways in which some of what we do is unlikely soon to be performed effectively by machines. But let’s reflect on what life may be like — for the average person — if legal assistance is ever as commonplace and inexpensive as refrigeration.

In other words, what if legal ‘ice’ became readily available on demand, with little cost? Will such an AI iceberg sink the legal profession? Will lawyers be able to spend more time reading Shakespeare? Will societies be more just?

Misdelivered

While writing this the following arrived in my inbox. It evidently was not to be opened until 2050, but somehow got sent back in time.

From: swonderstone@law.mit.edu
To: Undisclosed recipients — 5/16/2035, 9:34 AM
Subject: Don’t open until 2050

A View from 2035

Dear Future Friends:

Hello! If things go as planned, you’re opening this about 15 years after it was written. My assignment was to provide some color about the world of law way back in 2035. You’ll be glad to know that 2050, Here We Come is a current bestseller, written, like the last one, purely by machine, unstained by human editors.

It’s still early, of course. But things have begun to change. Here are some highlights.

The generative AI revolution of 2022–23 is now a distant memory. The hallucination problem was mostly resolved. Our latest machines don’t simply ‘say’ things, and tell us just about whatever we want to know, they can do things we want done. That includes ‘interviewing’ people, filing court pleadings, and interacting with other parties as an agent.

Thanks to Frontline Justice and related efforts, we now have a surplus of dedicated justice workers, some of whom are delighted to help people just for the opportunity of meaningful work, since so many other jobs have been taken over by machines. We’ve begun to talk of ‘employment as entertainment.’

Ever since LawWaze (“the sum of all legal wisdom”) and its many imitators went online in 2026, nearly everyone has had low-cost or free access to unlimited quantities of legal information and expertise, dispensed by emotionally intelligent, lifelike avatars. It’s almost like having a full time lawyer at your beck and call. Most people can also enlist a personal soldier from the army of human justice workers.

Of course the bar associations were up in arms, claiming the unauthorized practice of law, by both machines and unlicensed people.

The UPL wars finally ended in 2031, when a narrowly divided US Supreme Court decided, 8-to-7, that online legal advice systems were works of authorship, protected by rights of free expression under the First Amendment, and could not be suppressed by bar authorities as unauthorized ‘practice.’ Human advice giving, surprisingly, was likewise protected.

The court has also been experimenting with ‘machine hours’ as part of oral argument sessions, when litigant bots are allowed to present and answer questions. Many other forums now allow machine representation.

In 2032 states began enacting laws that established a ‘system-user’ privilege, akin to the lawyer-client and priest-penitent ones, for candid consultations with legal and medical advice systems.

Despite these mostly encouraging developments, gross disparities in outcomes persist. Some help is reserved for those willing to pay a premium. More generally, our societies still feature obscene inequalities in wealth and other advantages. And machine assistance only goes so far.

Looking back, commentators earlier in the century anticipated some of this. E.g.,

Some tend to regard the rise of machine intelligence as AI versus us humans. It will more likely pit humans against humans over access to such intelligence. It may be that one day only the rich will afford the best legal machine intelligence, and others will have to settle for flesh-based lawyers.
(From Toward a Phenomenology of Machine-Assisted Legal Work)

…

Even after AI and other automation and standardization have done their jobs we’ll still have the underlying challenge of helping people choose which contracts to propose and accept. Lawyers can’t fully answer those questions for business people or consumers. Nor can machines. Both can tell us what we should care about. But so far only people know (or think they know) what they do.
(From Shining More Light on Cares in Contracting)

While lack of access to legal assistance is receding as a social problem, our grandest hopes haven’t yet been realized. Systemic inequalities persist. People remain lazy, impulsive, greedy, and gullible. Justice is still unevenly distributed.

But I’m sure that will all have been worked out by the time you read this! In the meantime, I’m looking forward to the 26th International Conference on Artificial Intelligence and Law next year in Kyiv.

— Synthia Wonderstone, J.D. Candidate 2036, MIT School of Law

Looking Further Out

This unexpected glimpse into the future got me thinking. Can we extrapolate these trends? To pick up on some of Synthia’s points, do uncomfortable surprises lie ahead?

Back here in the 2020s, there’s obviously room for improvement. By any definition, blatant acts of injustice happen all the time. Many could be prevented or ameliorated with basic help. Even the powerful could be held to account. But dramatically more funding for free legal aid attorneys is not on the horizon. And small businesses and nonprofits also suffer from the lack of affordable legal assistance.

A common conceit among access-to-justice reformers is that radically more help in the form of human services is the solution. (If only everyone had a lawyer! A personal Perry Mason.) Another common conceit is that AI will come to the rescue.

Suppose that legal expertise and assistance — both machine-based and human — became too cheap to meter, and everyone had access to them 24/7. In other words, a nearly perfect legal help system. Could that really happen? What if it did?

Two scenarios

There are plenty of utopian and dystopian possibilities as to how such a ‘helpful’ world would be. Let’s first sketch a happy scenario.

Field of dreams

If the recent breakthroughs in generative AI are an indication, we can expect stunning improvements in machine capacity. Systems will fluently handle most human languages, written and spoken. They will dispense diagrams and other visualizations when needed to assist human comprehension. They will be aware of their own limitations, and show considerable emotional intelligence, presenting as photorealistic humanoid avatars.

Our future legal advice systems will excel in eliciting coherent narratives from people, leveraging new forms of artificial curiosity. Neuro-symbolic artificial minds will preside over vast federations of public and private knowledge stores. They will be able to review and assess every possible argument, counterargument, and stratagem in a given legal context. They will reliably predict outcomes.

Meanwhile, much of humanity will have been liberated from ‘routine’ work, and millions will have found meaning in helping others, including with legal problems and opportunities. Collection agencies and landlords will need to think twice before bringing legally dubious claims. Justice will flourish.

But there are less happy possibilities.

Stuff of nightmares

Lots of things of course could go wrong. Let’s acknowledge the potential for nightmarish scenarios — deep fakes, unchecked bias, algorithmic manipulation. We’ll likely see environmental impacts and other externalities of power-thirsty server farms and exploitative labor practices. Predictive analytics and facial recognition will be weaponized against the economically disadvantaged.

Billions of people may still not be online. Primary law may still be fee walled.

Systems may be controlled at various levels by humans with undemocratic socio-political preferences. Some will be tempted to abuse that power. Will authoritarian regimes even allow access to neutral legal assistance?

Improved visibility into the power dynamics underlying our legal system won’t change those dynamics. A stratified social system requires scarcity to maintain. Inequality will be perpetuated, if not exacerbated. Racism, sexism, ableism, and other maladies will persist. So might UPL regimes.

No collection of mixed metaphors about ice is complete without an homage to Ice-nine (a fictional polymorph of water in Kurt Vonnegut’s novel, Cat’s Cradle), which froze solid any liquid water it came into contact with (including eventually all of the world’s oceans.) Low-cost legal help could usher in an explosion of justiciable controversies, a hyper litigious era. Will we expand judicial capacity to rescue a system drowning in AI equipped litigants? Robo judges will introduce their own opportunities for mischief.

Complications

Time is the only dimension in which we’re all going the same way at the same speed. The road ahead for law and justice is ‘there,’ even if we can only see it dimly.

History will likely settle for a rough compromise between extreme scenarios like those above. Vastly more legal help will likely be available, but serious injustices will continue.

Put aside for the moment the nagging question of how a “benevolent omniscience on tap” and an abundance of human help might be funded. Consider, rather, some ways in which even the most advanced legal help delivery system — and a progressive social-political context — might fail to deliver the promised land some of us hope they would. What factors will likely play a role?

We can look at that from three overlapping perspectives, that of individuals, society, and technology.

Technology

We should expect continued limits in machine capacity, as relates to legal work. Some services may remain undoable mechanically, like ‘true’ counseling and zealous advocacy. Effective representation in unscripted litigation and negotiation scenarios may well elude machines past 2050.

Plus there are fundamental limits of artificial cognition when it comes to choices and judgments, which inescapably involve human cares and concerns that can’t fully be predicted or calculated. Judges, officials, and counterparties will often reach different legal conclusions, so the cognitive challenge ascends to effectively managing these differences. If price is no longer an obstacle, second and third opinions will often be sought.

Coordination challenges will continue to be presented when multiple people have to agree on a legal strategy or position. Factual investigation will still often require expeditions into the messy ‘real’ world.

Society

Technological revolutions can unleash disruptive reallocations of efforts and rewards. Law is a competitive sport, so a global injection of new resources is unlikely to change its core dynamics.

We should avoid confusing knowledge with power. Systems supply knowledge and affordances; they don’t supply connections or influence.

The Individual

No amount of technological progress will obviate the constraints of human nature.

People may be more aware of their rights and responsibilities, with ready access to legal information and expert advice. But for effective advice, the advisor needs to know more about the advisee’s personal circumstances than the advisee may care to disclose. It may require the advisee to do things she may prefer not to do, or to make choices she’s not prepared to make. People need to be willing to think things through; to plan, pay attention, and exert effort.

A chasm often lurks between intentions and accomplishments. Between learning what needs to be done to accomplish a legal goal and getting it done. Wishing something would go away and being willing to devote personal energy are different things. Some don’t want to bother.

Human attention and virtue will remain scarce; suspicion, resentment, angst, and other grievances will persist. Understanding and patience will remain in short supply.

Some will be unable to use these new resources. Language barriers, and perceptual, emotional, or cognitive disabilities, will stand in the way. Others may be unwilling to do even the rudimentary work necessary for someone else to effectively help them. And if you’re homeless or starving, ‘justice’ is of secondary interest.

Even if involuntary legal helplessness is eradicated, the disinclination or inability of folks to take advantage of help will remain. (Consider that even when highly nutritious groceries are readily available, many people still consume more junk food than they should. Folks will keep getting into the same kind of trouble, like lung cancer patients who insist on continuing to smoke.)

Residuals

What work will be left for human helpers? At least three things:

• Working with those for whom machine assistance is not feasible

• Acting as an agent or fiduciary for those who can’t fully handle their own affairs

• Supervising and training machines

What work will be left for the helpees? Again, at least three things:

• Providing information (about their circumstances, goals, preferences)

• Making choices

• Dealing with socioeconomic realities

We should pay attention, then, to the ‘receiving’ side of legal assistance, even as we attack today’s unacceptable shortages on the ‘giving’ side. How can we, for instance, do more to help those needing help make the best use of such help as they can get? Legal educators have a role in figuring that out.

The future is fluid

Ice ponds and wagons are things of the past Many of us now have ice cubes on demand, often via a dispenser on a refrigerator door. Cool drinks, at least, are much easier to arrange.

Modern life has seen many such transformations. Few ice block deliverers would have imagined that most of us would now have devices in our pockets through which we can communicate instantly with anyone in the world.

Law of course is different from frozen water. The analogy is imperfect. But legal help on tap could well become a new convenience.

Justice won’t roll down like water in a world of bottomless legal assistance. Abundant legal help of good quality won’t solve all our problems. Even complete knowledge and expert caring assistance won’t ensure fair outcomes or optimal legal wellness.

Let’s fight for them anyway.

— -

This piece began life as a talk at the first conference of the Future of Law Association in Dublin, June 2023.