We live in an age of hyper specialization. This is a trend that’s been evolving for centuries. In his seminal work, The Wealth of Nations (written within months of the signing of the Declaration of Independence), Adam Smith observed that economic growth was primarily driven by specialization and division of labor. And specialization has been a hallmark of computing technology since its inception. Until now. Artificial intelligence (AI) has begun to alter, even reverse, this evolution.
The ascendancy of AI has rekindled debate about the long-term capability of computers to either augment what we do or eventually replace us entirely. Highly specialized computing functions—designed to solve decision-making, learning-based problems like playing chess or filtering spam—are referred to as narrow AI. But AI that can adapt and respond to a broad range of external stimuli, illustrating a shift away from specialization, is referred to as artificial general intelligence (AGI) or strong AI.
AGI has captured the imagination of leading tech innovators and attracted massive capital investment. You can think of this as attempting to create software in our own image or at least how we perceive our cognitive functions to work. At the extreme end of the utopian or dystopian spectrum, depending on your point of view, is super AI or artificial super intelligence, the most generalized form of AI, that would, theoretically, have abilities beyond that of humans—which could either propel us into a contented life of leisurely pursuits or turn us into a super AI server servant class.
The ultimate aspiration of much high-profile AI innovation is to move away from hyper specialization and become generalized. A computer may be able to beat the greatest chess player in the world, but can it function out in the world even at the level of your dog or cat?
Doubtful at the moment. But the direction is clear. While society is moving toward ever more specialization, AI is moving in the opposite direction and attempting to replicate our greatest evolutionary advantage—adaptability.
Working in an AI World
Measuring AI’s potential for job replacement is difficult, but there has been some analysis. The World Economic Forum predicts that AI will replace 85 million jobs by 2025. A 2023 study from Resume Builder suggests that more than a third of businesses have already been replacing workers with AI. Well-known industry leaders and experts tend go even further—Elon Musk has stated that he believes most jobs will eventually be replaced by AI while Open AI CEO Sam Altman takes a more moderate view that all repetitive human work not requiring a “deep emotional connection” between people will be done better, cheaper, faster by AI.
What does this mean for the future state of work in an AGI world? That depends on how you view what we actually do for a living. Let’s look at one prevalent profession in the U.S., some might argue an overrepresented one: lawyers.
The ABA estimated the number of U.S. lawyers in 2023 to be about 1.3 million. How much of a lawyer’s work could be subject to replacement by some form of AI, whether narrow, strong or super? Are they consummate creative beings far beyond the capabilities of even advanced generative pretrained transformer (GPT) miracles?
In The Wealth of Nations, Adam Smith famously analyzed the steps required to make a pin, which he concluded required 18 different specialized tasks. One worker could take a whole day or more to make a single complete pin. But dividing the work into individual tasks among 10 people, he asserted, could produce more than 48,000 pins per day.
How Much Can AI Take Over?
Putting aside the question of whether making pins is more intellectually stimulating than reviewing discovery documents, the inherent set of document review tasks seems to be tailor-made for narrow AI. Analytical tools can identify authors of documents, parties to conversations, dates of transmission and receipt. These are the type of clearly defined, specialized tasks that computers have been great at practically since their inception and at which narrow AI excels.
But could AI take over the bulk of legal work or is there an underlying thread of creativity and judgment of the type only speculative super AI could hope to tackle? Put another way, where do we draw the line between general and specific tasks we perform? How good is AI at analyzing the merits of a case or determining the usefulness of a specific document and how it fits into a plausible legal argument? For now, I would argue, we are not even close.
Whether considering discovery document review, drafting contracts, legal research, compliance monitoring, or analyzing litigation viability we currently live in a narrow AI world. Lawyering, particularly specialized tasks, can be greatly aided by AI but at present is not about to be fully replaced by AI however you classify it. If you’re waiting for that you’re just going to have to put a pin in it.
