Author: 0xAlpha
(Originally published on 2019-05-06, translated into English on 2022-10-29)
At this historic moment of 2019, the two most interesting things that will happen to human society in the next two decades are basically clear.
One is AI, and the other is crypto and blockchain.
What's even more interesting is that these two technologies will grow into two opposing forces at the ideological level. This could be one of the ultimate battles of human society in the next decade.
As the popularity of AI continues to rise in recent years, there have been a lot of discussions about its impact on human society. Some of them are fears of human-AI confrontations similar to that in "Blade Runner" or "Westworld", which originated from the AI fantasy in the literary and artistic fields. Whereas other discussions just regard AI as something similar to steam engines, electricity, and the Internet–nothing but another tool upgrade. The truly profound impact of AI has rarely been discussed.
The most topical scenario, the awakening of AI self-consciousness as in "Westworld" or "Blade Runner", will not happen in the foreseeable future because AI is very, very, very far from autonomous consciousness. However, that is not to say AI won't make a disruptive impact. The most significant change brought about by AI is that in the future, human beings could significantly reduce their dependence on the element of "human being". This is the fundamental change driving the disruptive impact of AI in many fields. My article will make linear extrapolations of the AI impact in three areas. However, the nonlinear effects brought about by AI are still difficult to imagine.
Before the discussion, let's get on the same page about the definition: AI discussed in this article is a general term referring to machines (including computer programs) replacing the role of humans in work and requiring little human guidance or intervention in the process. This definition actually includes part of the field of automation. For example, the early robotic vacuum cleaners (e.g., the random-walk models of iRobot) do not use artificial intelligence algorithms in the narrow sense. However, they still fall under the general AI umbrella in my discussion. Not surprisingly, the boundaries between AI and automation are inherently blurred.
I will discuss three areas AI would impact: economics, military, and politics.
Let me start by throwing out my prediction: the mass adoption of AI technologies will significantly exacerbate the wealth inequality of society. The wealth-gap-widening effect of AI will surpass any other technology in history. This is because although most people may enjoy the technological benefits of AI (enjoying the services or products provided by AI), only a tiny fraction of people can enjoy its economic benefits (making money from AI).
With the AI era in full swing, there are four categories of people, divided by the extent to which they will be affected by AI.
AI investors, AI entrepreneurs, and owners of AI core technologies: overall, this group will enjoy the AI economic benefits exceeding that brought by the Internet to their internet counterparts (i.e., internet investors/entrepreneurs/core technology owners).
AI workers: AI will establish a new industry and hire workers. But the demand may not be as large as that of traditional industries like manufacturing. AI workers are also beneficiaries of AI technology (after all, they get paychecks because of AI). But their share of economic benefits may be relatively small, probably equivalent to the economic benefits of programmers in the Internet industry. Actually, the threshold for becoming an AI worker isn't that high. In the past ten years, the computer industry has dramatically lowered the threshold for using AI technology: you don't really need to have a deep understanding of things like random forest, support vector machine, or convolutional neural network. Just having learned some programming basics and installed sklearn or TensorFlow on your computer would get you started as an AI worker.
AI-neutral people whose jobs will neither benefit from nor be harmed by AI, including artists, politicians, lawyers, and other professions that AI technology cannot cover.
People that will be partially or entirely replaced by AI in their jobs: assembly line workers, restaurant servers, cashiers, cleaners, etc. The impact on this population segment is difficult to predict precisely but is likely to go beyond the expectation of most of us.
Since the role of AI at work directly overlaps with that of people, its impact on the micro-level of economics may exceed that of any previous technological revolution. Some would argue that any technological revolution creates a new group of jobs while decimating an old one. For example, the steam engine put horsemen out of work but created the profession of drivers, the latter a much larger employment opportunity than the former. However, this time is different! The previous technological revolution went like this: the capitalists were dissatisfied with horses, so they replaced horses with steam locomotives, and later dissatisfied with steam locomotives, so they replaced steam locomotives with petrol engine cars. However, this time the capitalists are dissatisfied with the drivers themselves and will replace the cars with driverless ones! Do you get that difference? The previous technological revolutions were all about tool upgrading–they replaced old tools with new ones. Accordingly, the people operating the old tools are replaced by those operating the new ones. But this time, the capitalists are dissatisfied with the workers themselves: "They are lazy and emotional. The unions tell me every day that they can't be fired and the benefits can't be reduced. Sometimes they even go on strikes! OK, fine, I am done with this game. Let's stop complaining about the long working hours. How about we get some AI robots to work 24-7?"
Let's look at an example showing why AI is the ideal worker. The manufacturing giant Foxconn probably represented the highest level of human manufacturing efficiency. What was at the core of Foxconn's management model? It was really about how to make workers work like robots. Nevertheless, workers can't be more robot-like than real robots by any means. For example, the worker suicide problem is probably an insoluble one for Foxconn. But AI will never kill itself. Hence AI is indeed the ideal worker for capitalists.
Almost every technological change in history has led to some degree of wealth differentiation. But previously, as the business grew bigger, capitalists and entrepreneurs had to employ more people and share the economic benefits of that technology with them. For example, the Alibaba IPO not only created the richest man in China and the richest man in Japan, but it also created dozens of billionaires and many more multimillionaires. As of today, thanks to the employee stock ownership plan, most of Alibaba's 60,000+ employees can share the economic benefits of e-commerce technology. Whereas in the case of AI, while the capitalists and entrepreneurs pioneering the use of AI would enjoy substantial economic benefits, they don't have to employ so many people as in traditional manufacturing or internet industries and thus don't have to share the economic benefits with a large group of people. If you have seen Tesla's inhuman factory, you know what I am talking about.