The age of the generalist

A polymath on a computer imagining the future. Generated by DALL-E

Intro

Knowledge workers have existed all throughout history. From Plato to a modern developer, people who have focused their lives work on producing knowledge artifacts have existed and have contributed to our society. With the emergence of AI assistants, knowledge work will change and I suspect it will be revolutionized. Lately we have experienced that specific knowledge is less necessary, however having a global understanding of different areas of knowledge is key. Memorizing computer commands, formulae, words or even small facts is absolutely unnecessary. Everything is just a prompt or search term away.

This week we had a very interesting experience at work. After a bug we needed to fix some records that got calculated wrongly. We started extracting some information and comparing them to an output of internal apis to identify the affected transactions. I, being a nerd as I am, gave some precise prompt and uploaded a single file to the code interpreter which gave me an answer in about 30 seconds.It basically reproduced all calculations that our API did. A senior dev did the same exercise but building as script and obtaining results manually. The idea is that we could crosscheck both our work and see if both matched. The LLM result was our sanity checks. I came back about 1.5 hours later and the result was the same, exactly the same.

I don't expect somebody without prior knowledge can come and build the prompt and extract the data as I did. That takes specific knowledge on the task at hand but comparing myself to the senior dev, he is by far more skilled than me. He uses vim as his main IDE, so yeah, he´s hardcore. The question is, how could I, a generalist product manager, match up to a senior dev in skill using an LLM and beat him by hours? This was not possible a few years back but it is now.

The archetype of a generalist

The archetypal idea of a generalist is a polymath, an individual whose expertise spans a myriad of different subject areas. Such a person is known to draw on complex bodies of knowledge to solve specific problems. The term "polymath" is often used to describe great thinkers of the Renaissance, such as Leonardo da Vinci, who excelled in various fields including art, science, mathematics, engineering, and many others. A polymath is a figure who embodies the ideal of human potential and the limitless capacity for learning and understanding. It represents enlightenment and the pursuit of knowledge, signifying the idea that one should strive to grasp the vast complexities of the world. The polymath's insatiable curiosity and ability to master diverse fields of study can inspire awe, admiration, and sometimes even a sense of mystery or otherworldliness.

How does the perfect generalist look like

John von Neumann (1903–1957) was a Hungarian-American mathematician, physicist, computer scientist, and true polymath. His contributions spanned various fields, including quantum mechanics, game theory, economics, and computer science. He is often said to be one of the most intelligent individuals to have ever lived.

One of the most astonishing aspects of von Neumann's intellect was his ability to perform complex mental calculations. As a child, he could divide two eight-digit numbers in his head and write impossible mathematical proofs minutes after the problem was presented to him. Von Neumann was sent to study calculus with the famous Gábor Szegő, who was so astounded with the boy’s talent that he cried on their first meeting. Von Neumann's work defined modern computer architecture, which is known as the von Neumann architecture and is the basis for most computer systems today. His vision and understanding of both hardware and software were prophetic. In addition to his technical prowess, von Neumann was fluent in several languages and had a deep understanding of various cultures and their history.

John von Neumann's life is filled with stories that illustrate his unparalleled intellect and wide-ranging contributions to science and humanity. Whether it was his rapid assimilation of information, with the ability to read a book in one glance and retain its contents, or his groundbreaking work in game theory that laid the foundation for a new field of mathematics, von Neumann's polymathic abilities continue to inspire awe.

How is a generalist different

Those who met Von Neumann believed that his mind represented an evolution of the typical human intellect. What if AI and modern search were augmentations for our brain’s power that made us reduce the distance to someone like Von Neumann. We live in a world where the capacities of this marvelous man are easier to attain.

What made Von Neumann different? I’ve compiled a few traits which I think made Von Neumann a superhuman. I’ll list them and I´ll try to discuss how AI can help us reach them?

Mental Calculation - Von Neumann could perform complex mathematical calculations in his head with astonishing speed and accuracy. In the modern world, especially after the personal calculator revolution of the 60s, mental calculations are a thing we don't have to worry about. It is to become a lost skill. In the right hands, scientific calculators could also surpass the capacities of humans and it has become a walk in the park for something like Code Interpreter paired with a knowledge database like wolfram AI. I believe modern humans should still learn math, especially understanding its underlying principles.

Photographic Memory - Von Neumann reportedly had the ability to glance at a book or document and recall its contents in detail, a trait often described as a photographic or eidetic memory. With things like OCR, modern technology will allow us to literally have this ability. That aside, an LLM is right now trained on trillions of tokens which will aid us in remembering long forgotten knowledge. I think we could do more in this aspect in our daily life. We have the capacity right now of saving all data we learn, even by just saving our digital notes or emails. This is an amazing ability that I think many of us underutilize. That said, good memory is a big advantage to any brain, even if you use an AI system, being able to recall long forgotten knowledge to write the perfect prompt is still one of those key abilities.

Multilingualism - Von Neumann was fluent in several languages, allowing him to communicate and collaborate with scientists and scholars around the world. I think we are not that far away from a real time translator, from what I´ve seen lately. Even then, English has become kind of a lingua franca, much like what Latin was in the middle ages. All that said, multilingualism seems to be beneficial for the development of better intelligence, it brings cognitive flexibility, enhances attention, develops memory, improves cultural understanding and improves problem solving abilities. That in itself makes new languages key for intelligence development.

Interdisciplinary Genius - Von Neumann's intellect was not confined to a single field. He made significant contributions to mathematics, physics, economics, computer science, and more, demonstrating a rare ability to master and innovate across disciplines. This is probably the most difficult for humans to attain but easier for large language models to provide. The breadth of knowledge that an LLM brings, provided by the sheer amount of tokens it’s being trained on gives it the depth that we lack, just because of our limited capacity to read. It is said that a human will not read a billion tokens in his/her lifetime, modern language models are easily over an order of magnitude bigger. Having an interdisciplinary mind will fit perfectly to the use of LLMs. If you are one of those generalists, celebrate because this is your era.

Problem-Solving Skills - Von Neumann had an uncanny ability to analyze and solve complex problems, often finding solutions that eluded other brilliant minds. I think this is where it would be most difficult for an AI to help us. Most of the problems that the AI can encounter from a human prompt, it has already seen. For those problems AI will work great, but where it will falter is when it encounters a new problem. In order for AI to solve problems we need to get closer to AGI (artificial general intelligence). Right now we are using only symbolic AI and building neuro-symbolic systems. This has been path dependent on how we have developed AI. In this area, humans remain unmatched.

Intuitive Understanding of Abstract Concepts - Von Neumann's grasp of highly abstract mathematical and scientific concepts was profound, allowing him to develop theories and models that have had a lasting impact on various fields. AI is extremely helpful, it knows associations nobody else knows.It seems to hold logical rules expressed in language that we have not seen, just based on the sum of the probabilities. Given enough data AI will be able to learn a language, understand its semantics and abstraction as well as any human being. We need to know the language very well too in order to find patterns within the AI. This is something that needs to be studied in depth.

Creative Thinking - Von Neumann's creativity in approaching scientific and mathematical problems led to groundbreaking work in areas such as game theory and quantum mechanics. True creativity, true production of new knowledge is where I think humans shine. Especially in math, the list of unsolved problems is really long. Until AI reduces that number we can safely say that creative thinking is still our most amazing advantage. Creativity is not about producing images or songs, it’s about finding new ideas and knowledge. Until through AGI appears we are still way ahead of modern systems.

Wrapping up

In the era of rapidly advancing artificial intelligence, the archetype of the generalist and the incredible capabilities of figures like Von Neumann intersect with the potential of modern tools like LLMs. The convergence of deep interdisciplinary knowledge and the augmentation of AI presents a new frontier for us knowledge workers. No longer bound by the constraints of specialized expertise, generalists are now equipped with AI tools that can amplify their wide-ranging knowledge and intuition. While certain human faculties like creativity, problem-solving, and deep understanding of abstract concepts remain unparalleled, the partnership of generalists and AI is reshaping how we approach knowledge work. We are living in times of transformation, it's imperative to harness both the power of human potential and technological advancement. We are witnessing a renaissance of the generalist, where the synergy of intelligence and machine elevates our collective capabilities. I hope this brings us a new era of enlightenment, much like what transpired approximately 500 years ago..