Elastic Intelligence: Beyond Bodies, Bounded Rationality, and Billable Hours
Where I explore how AI frees intelligence from the limits of population and time
This is the third article I’m publishing under Deep Legal Consulting.
I recently presented my thoughts on transforming legal service delivery with AI to a consulting client. And, it got me to thinking about how our understanding of today’s technology can limit what we believe we can achieve next year or the year after.
In this article, I attempt to suss out a different mental model for how to think about AI. I hope this one resonates with you, my dear readers. Enjoy!
A familiar truth in law practice is that some matters are won by muscle and some by insight. If you have a document dump the size of a warehouse, you staff it. If you have a knotty jurisdictional question that turns on a single sentence in an old circuit split, you call the partner who can see around corners. We have always scaled the first with people and the second with rare talent. That habit was born of necessity.
And here is the opportunity of our moment.
Artificial intelligence moves cognition onto a substrate that is not bound to population or the clock. Intelligence begins to ride on electricity. The unit of thought is no longer a person. It is a process that can be copied, scheduled, and orchestrated. For lawyers who have lived inside the limits of headcount and billable hours, this shift is as consequential as the move from paper to digital discovery. It does not just speed the old workflows. It changes what is thinkable.
If this sounds interesting to you, please read on…
This substack, LawDroid Manifesto, is here to keep you in the loop about the intersection of AI and the law. Please share this article with your friends and colleagues and remember to tell me what you think in the comments below.
Population Used To Be Our Destiny
For centuries, the way to apply more intelligence to a problem was to add people. Cities that gathered larger populations also gathered more specialists. Universities grew in breadth as they grew in headcount. Firms scaled by training classes of associates. The shared assumption was simple. More minds meant more thinking.
Human thinking power has been proportional to the number of trained minds and the hours they can work. The industrial era paired brains with the logic of the assembly line. Break the task into parts. Assign each part to a team. Move the work forward in parallel. Legal operations borrowed the same logic of billable hours and assembly line efficiency with the help of Reginald Heber Smith and the Cravath system.
Biology set a hard ceiling. Each person has a single stream of attention (although some claim to be able to multi-task), a need for sleep, and a limited working memory.
We built institutions that compensate for these limitations. Hierarchies aggregated decisions. Committees reduced noise. Precedent compressed the search space by telling us what had already been tried. The system worked well when the bottlenecks were paper, shipping, and meeting rooms. It faltered whenever the problem required a new mindset or theory rather than more labor.
Genius As Biological Bottleneck
Geniuses do not grow in a vacuum. They emerge from particular environments that meet basic human needs and then layer on access to education, libraries, mentors, and time. Plenty of bright minds never get that combination. Others have it all and still do not catch the spark. The result is a spectrum of intelligences in which the rarest contributions depend on both ability and circumstance.
We often speak of brilliance as if it sits atop a ladder. In reality it is a bottleneck at the narrow end of a funnel. Many competent practitioners can execute a known playbook. Fewer can write a new one. The conditions that produce a Newton or an Einstein have to line up just so. Intelligence must be sufficient. Hunger and motivation must be present. The environment must keep the person fed, safe, and curious. The right books must be within reach. The result depends on inborn aptitude and on luck.
Herbert Simon, whose work on “bounded rationality” reshaped economics and social science, taught that humans settle for good enough, rather than perfection, because attention and memory are scarce. We pick good enough options because our brains are finite. That observation explains why change is hard and why step changes are rare. When a new concept appears, it unlocks a fresh branch of the tree of ideas, and a lot of mental labor to accommodate the change.
When Intelligence Rides On Electricity
Artificial intelligence shifts the production function. Instead of hiring more people to add minds, we can spin up more processes. An AI agent can be copied in milliseconds. Copies can work in parallel, pause when idle, resume on demand, and be coordinated by a scheduler that never tires. The inputs are computation and power rather than years of schooling. We still need humans to design goals and judge outputs, but the doing of the thinking scales in a new way.
This is not science fiction. Even without AGI, narrow and multi-domain systems already handle tasks that once required entire teams. Summarization at scale. Pattern finding across millions of documents. Drafting that respects constraints. Mathematical search over high dimensional spaces. If we assume further progress toward more general reasoning, the set of tasks that can be parallelized expands. The constraint becomes orchestration and strategy, rather than headcount.
Intelligence becomes elastic.
For lawyers, the analogy is familiar. Think of AI as a co-counsel who can split into a thousand competent clerks, each taking a narrow slice of inquiry, then reconvene as one voice. That is not a replacement for judgment. It is a multiplier for it. The lawyer becomes the architect of questions, the curator of answers, and the guardian of risk.
Finesse and Brute Force
Legal work has two tempos. One is brute force. The other is conceptual. AI affects both. On the brute force side, agents can create exhaustive maps of fact patterns across jurisdictions, track how a term of art or legal concept drifts over time, or simulate how a clause performs under different factual assumptions. On the conceptual side, agents can generate competing theories of the case and stress test them against the record. They can finesse edge conditions that even careful teams miss because the possibilities loom too large.
This is not a claim that machines will have intuition in the human sense. It is an acknowledgment that machines can flood a problem with different approaches and test them in parallel. Picture a moot court with a thousand competent debaters, each pressing a different angle. Now picture your ability to choose the five strongest threads, refine them, and present a narrative that aligns with the forum and the facts.
The Discovery Function
I.J. Good described the idea of an intelligence explosion in which a smarter system can design even smarter successors. We do not need to embrace the full arc of that argument to see an immediate effect that matters for professionals. When thinking processes can be run in parallel across massive search spaces, we discover more.
We do not only find more needles in haystacks. We find haystacks we did not know existed. The space of viable theories expands because more paths are tested. In a sense, this harkens back to Parkinson’s law: that work expands to fill the space of time allotted to it.
Discovery is not only about novelty. It is also about validation. The same elastic intelligence capacity that generates new approaches can be used to test them. In law and policy, that means fewer blind spots. It also means we can evaluate the second and third order effects of rules before they are enacted. When we model how incentives ripple through a system, we move from accidental wisdom to intentional design. Bounded rationality remains a fact of human life. Elastic intelligence becomes a tool that supplements it.
Solutions Reveal More Problems
Difficult domains often hide multi-stage gates. Like a video game, solve level one and the door to level two appears. Each solution exposes a higher resolution version of the next problem. Energy storage improvements create new opportunities in electric grid design. Grid design improvements shift the economics of desalination. Better desalination changes agricultural options and trade routes.
Elastic intelligence will not make scarcity vanish overnight. It will make more of these gates visible and passable because we can pursue many lines of innovation at once. It will be jagged gains, but these gains will compound across domains and open up new possibilities we never knew existed.
Preparing for Elastic Intelligence
Professor Richard Susskind challenges us to ask ourselves, “What if AGI?,” as a way of not giving too much weight to the technological limitations of the moment. How do we prepare for elastic intelligence?
Firms and legal departments will need to cultivate a new craft that sits between research and engineering. Some practitioners will specialize in problem decomposition, prompt specification, and evaluation. Others will become guardians of provenance and privacy. Training will blend doctrine with systems thinking. Work culture will prize curiosity about new technological techniques and tools, but it will also prize restraint about when to trust them.
Closing Thoughts
I began with a simple observation. Biological intelligence scales with population and time. That constraint gave us assembly lines for routine work and rare geniuses, like Einstein, for conceptual leaps. Artificial intelligence loosens that constraint by letting thinking processes live in a way that can be multiplied, coordinated and run in parallel.
As a lawyer, I am drawn to tools that improve foresight. Clients hire us to see the consequences that others miss. Elastic intelligence is a way to widen the aperture. It lets us test more paths in parallel, catch errors and failures before they metastasize, and support our theories with richer evidence. It can also tempt us to overreach. The remedy is the same as it has always been. Ask precise questions. Validate unusual answers. Keep a transparent record that earns trust.
The endgame is not automated law or automated governance. The endgame is a stable environment for human flourishing that stands on institutions that can better digest complexity, help us to see ahead, and succeed. If we do this well, we will remember the headcount era as we remember the first personal computer: grateful for what it enabled and relieved that intelligence is no longer bound to bodies, population size, or life span.
Freed from those limits, we can finally pursue advances that our own biology, luck and environment have kept out of reach: more minds than people working on every hard problem, more testing of solutions than hours in the day, and outcomes shaped by design instead of chance.
Now that is a future I can look forward to!
Genius level articulation of how we can leverage AI to higher potentials.