AI cannot replace the shared economy in the next few decades 2.0

Editor's note: This is an article published on the official website by Aleph, a famous Israeli venture-investment agency. Google has been developing unmanned cars, really to replace human intelligence with artificial intelligence, but this is a great challenge to AI itself, the price is very high. At the same time, many of the world's labor force is idle, including those who can be drivers. So, if we can mobilize the idle labor force, like Uber and other shared economic models, we do not need so high price to drive the car, you can find the driver in real time. We can call the Uber and Airbnb, such as the successful model of the "shared Economy 1.0", it turns out that the shared economy has a great advantage in mobilizing idle resources, but in the face of the unequal distribution of the global labor force, we also need to exert more creativity to open up the "shared Economy 2.0".

Artificial intelligence is not the creation of artificial science, it is not to understand the human intelligence of science, it can not even imitate human behavior, so that people think of machines as people, like Turing Test. Artificial intelligence is the science of creating machines that allow machines to accomplish tasks that humans can do or want to do.

Google cars, sensor networks and robots are all tied to machine vision and automated computer control, creating a good news story: People vs. machines, human intelligence vs. artificial intelligence. But I want to calm down and pour some cold water on the news story. Compared to artificial intelligence, there is a place where someone can do better and cost less, that is, "share Economy 2.0".

We can call the Uber and Airbnb, such as the successful model of the "shared Economy 1.0", it turns out that the shared economy has a great advantage in mobilizing idle resources, but in the face of the unequal distribution of the global labor force, we also need to exert more creativity to open up the "shared Economy 2.0", For example, TRIPMD, a start-up company in the United States, uses the shared model to help users find quality medical services outside the United States that are cheaper than local prices. You can also try to share patterns like "parking". For example, a residential area in a city, when he drove to work in the morning, his parking space out, at this time, the other happens to the surrounding people can park the car in the vacated position.

The perfect AI is hard to achieve. The human brain is the result of millions of of years of evolution, and it already has the inherent ability defined by rules that have not yet been clearly understood, which can handle the various situations that programmers call "extreme cases" (corner cases). This is because the human brain has some learning abilities that cannot be completely explained. AI does not allow any failure, and most software engineering is not perfect, as long as "good enough" on the line, as Zuckerberg's famous saying "completion is more important than perfection." This is possible for software, because the first 80% of the revenue from a software project costs 20%, but the remaining 20% of the revenue takes 200% of the cost and time.

This can be explained in economics. For example, there is a large number of drivers and security supplies, but they can not find where there is demand. And the global labor costs are unevenly distributed, as India's median annual income is $616 trillion, while some parts of the United States are $13794 trillion. Another trend sweeping the global economy is the increase in the number of remote workers, and a study by global Workplace Analytics shows that Home office numbers will increase by 63% over the next 5 years. There are so many cheap labor, there are more and more fast chain relay and remote technology, We can use the manpower to complete a lot of work with robots.

At the same time, sharing the economy is also reducing the cost of living. Each enterprise that does "share Economy 2.0" service is gradually realizing marginal optimization, and the automation part is also increasing. However, such services also have some obvious difficulties, such as the trustworthiness of mutual communication and the length of waiting time, as well as the cost of deploying remote equipment, and various regulatory issues, such as whether the service provider is qualified. These problems are not the same as the current Airbnb, Uber and so on.

Just as demand for big data drives the development and use of Hadoop, and makes AWS a money-making machine for Amazon, the platform for sharing the economy still needs to be developed. This requires a platform like the Amazon-developed "Turkish robot" that uses people's networks to perform tasks that are not suitable for computers, unite people with AI, and provide real-time services on a global basis. This will take decades to achieve. People will also be driving cars because humans are less expensive to drive cars for a long time to come, and humans are better at intuitive processing than AI when driving a car.