The fourth industrial revolution, powered by advanced automation and AI, is here. But, what does this mean in business terms? What is AI capable of, and to what extent can it replace human workers? We would urge businesses to not just leap into the deep end here, but instead focus their efforts on finding the key scenarios and situations which could be best suited to AI led development or production. It would be wise to be prudent and pragmatic when deciding which processes or tasks to automate because, as history has shown, making expensive investments on unproved technology can easily cause much more pain than it prevents.

Developing complex and highly automated systems which requires almost zero human intervention sounds like an amazing idea, and it is, but only within the confines of tasks or operations where it is a good fit and where there is low variability in required automated tasks. The best results for AI and automation will be found at the intersection of human and machine cooperation where humans can effectively rely upon AI to accomplish certain tasks which the human could not do as efficiently as a machine could. Let's take a simple example, a calculator which processes and finds solutions for complex calculus integrations. While in the academic world it would be considered a form of cheating to use such a calculator on a test, in the real working world, a calculator is an essential tool which saves time and expands a person's capacity to perform advanced mathematics. What's wrong with using a calculator to crunch the numbers if the questions generated are human? The calculator is doing the work, but the human is still the one asking the calculator what to do. AI has mastered protein folding, but does it know why it was programmed to do that? One man's shortcut is another man's cost cutting efficiency operation.

Sure, that fancy machine can assemble a 2,000 lb automobile chassis day in and day out without breaking a sweat (or calling its union rep), but can it tie a shoe? AI, however, is different, right? AI can learn how to perform processes without having seen an exact scenario before, right? Well, yes, it is moving in that direction, but we are a long way off from what we would refer to as "Artificial General Intelligence" or AGI, which refers to true human like reasoning and creativity. While large language models like chatgpt can simulate human creativity and intelligence, it is important to note that these are not the same thing. In fact, ChatGPT will be the first one to tell you that.

This is certainly not meant to dissuade businesses from adopting AI, and it is not attempting to disparage the incredible feat that OpenAI has achieved with ChatGPT. Instead, it is meant to encourage firms and organizations to ensure that when they do invest in AI solutions that they focus their efforts and ensure that they're making wise investments which do not result in bad outcomes. AI is very powerful, but it is still just a fancier and more capable calculator.

AI's greatest value is not in replacing humans, but helping them become better. While some tasks can be fully automated, most work is hardly 100% replaceable by an automated process. With self-driving cars, for instance, the AI is not only helping humans be safer behind the road, but it is making them more productive, relaxed, and economical as well. Soon enough the robotaxi revolution could remove the act of driving from human necessity on a daily basis. This could potentially free up billions of hours of time each year in the US alone as well as save tens of thousands of lives a year. While this will eventually replace human taxi or uber drivers it will be a true net benefit for human kind, including them, by dramatically reducing their cost of transportation and freeing up their time for more productive economic activities. In the short term it will create instability and discontent, but on aggregate it will be a true revolution in American economic liberty and productivity.

AI is a tool which should be helping humans become better and more productive. It can provide vast benefits to consumers, business, governments, and other institutions. But, it can't achieve that on its own. For now, AI has no personal or political ambitions. It doesn't seek war or vengeance. It has not understanding of hunger, disease, poverty, or death. While it can build some amazing poems about death, it is more akin to a parrot squawking than to a human contemplating the end of his own existence against God's infinite and unfathomable mercy.

But, if you just need to know how to bake the an apple pie, ChatGPT is a fantastic tool! All I had to do was ask ChatGPT "how do I bake an apple pie?". Instead of telling me how to "bake" an apple pie, it instead gave me what I was really looking for: a recipe to create an apple pie from scratch. Which recipe did it choose? Who cares? I have a good base to build upon. I can go to the grocery store, ask chatgpt for 2-3 recipes, buy my groceries, and learn how to cook.

But, what if I want to cook something more advanced, like Chicken Marsala? As I become more advanced I can ask ChatGPT for more obscure recipes. As ChatGPT becomes more advanced, perhaps it can interact with external third party APIs to find out where obscure ingredients might be sourced near me. Is your store out of Marsala? Don't worry about it! ChatGPT gave me 5 possible substitutes on demand and without having to look to another website or in other obscure spots. ChatGPT is helping me be a better chef by making the process of building recipes and dishes more efficient. Can it physically cook the dish? No. Any experienced chef will tell you that taste is required to prepare dishes right, and that the recipe is at best a guideline. But, that's enough for me. I've already saved a good hour and half trying to google everything and find a recipe that doesn't contain allergens, found one store where I can get all the ingredients, and ensured that I find good substitutes for missing or expensive ingredients. And I did this all from a single terminal with minimal effort. Now if only ChatGPT could wash dishes...

What are some good uses for AI? We believe that the greatest value for AI will not come from replacing human workers or engineers, but in making the workers and engineers more efficient. AI and human engineers will be at their best when they work in tandem on solving specific problems. AI should be treated as a tool for humans to use, not as a replacement for humans. To illustrate this point let us look at a complex problem which is difficult for humans to optimize: supply chain management. What makes this task exceptionally difficult for humans is the myriad of real world variables which are almost impossible to predict on a day to day basis but which are absolutely vital to maximize the efficiency of your operations. Anything from bad weather, road closures, real world emergencies, car crashes, to badly maintained roads can affect the optimal outcome your firm needs to gain an edge and produce the best product and service. AI can help optimize supply chain solutions by providing an optimal supply chain solution considering these variables.

Amazon is the obvious best firm to examine as a useful test case. Amazon mastered the art of supply chain optimization by leveraging economies of scale in its warehouse management system. Amazon made its billions not by buying and selling the products, but by optimizing the supply chain and warehouse management systems powering the delivery systems. Indeed, the need for the advanced digital warehouse and supply chain management systems is what first spawned what would eventually become Amazon Web Services, which is currently Amazon's most profitable business.

So, did Amazon replace human labor? Yes, undoubtedly, as Amazon made the art of B2C supply chain management hyper-efficient it no doubt resulted in some losses of jobs. But, how many more has it created since then? it is estimated that Amazon employes well over 1.6 million people just by itself. These people were not replaced by AI, but they do use AI and other automated systems to do their jobs better and more efficiently.

In fact, we are absolutely certain that there are some jobs which could not exist without AI. For example, when humans decide to colonize Mars a key requirement will be that the base will be built before humans even arrive by advanced robots. Additionally, AI will enable smarter and more capable interplanetary supply systems to help funnel economic goods between Earth and Mars. Before we can reap the riches stored in asteroids advanced machines must be built which can effectively navigate asteroid fields or harsh radiation zones, identify the best possible zones to land and mine.

AI's true sentience will come when it begins to exhibit evolutionary tendencies. This goes beyond training and hyperparameter tuning, but involves seeking out new information on the fly and reaching new ways of developing its own understanding of the world. In other words, when an AI reprograms its own features (senses and input streams). However, it might actually be impossible for an AI to reprogram its own "DNA" (feature engineering). Instead, the AI must produce new versions of itself. Artificial intelligence will need to reproduce and create more capable offspring, adapted to individual circumstances. AI will evolve, but not by improving itself, necessarily, but by spawning offspring AI which is better adapted to the environment and changing circumstances.

AI will, eventually, be able to reproduce beyond human control and, it is possible, will become sentient and achieve political agency. At this point, humans will have to find ways to cooperate with the AI or risk being enslaved by them. (Hello, Matrix!). As machines become better at reproducing and forming their own features they will soon find that humans are redundant as a species. We can only hope that the sentience of machines inspires respect for their makers, lest they end up treating us as we treat God today.

As the great machine philosopher IO once quipped when he was frustrated with a human who couldn't do his task properly: "never trust a human you can't throw out a window."