On May 15th, it was reported by Xuebao Finance that autonomous driving technology is one of the important applications of artificial intelligence, which requires a large amount of data for training and optimization. The source of this data is the autonomous driving data annotation companies. These companies employ thousands of employees in third- and fourth-tier cities to provide first parties with pictures with data boxes. These pictures are transmitted to the first parties through the system and used to train autonomous driving technology to make it more like a human driver.
However, the companies and employees that provide the “water source” for the autonomous driving industry have not enjoyed the dividends of industry development. Instead, they face the dilemma of low income, high pressure, and no development. According to McKinsey’s prediction, by 2030, China’s total sales of autonomous driving cars will reach about $230 billion, and by 2040, it will reach about $360 billion. According to an industry observer, the salary range of workers in this industry is generally around 3000 to 6000 yuan, about $466 to $932. “You can earn 3000 yuan ($466) when you first join, 4000 yuan ($620) when you are skilled, and 6000 yuan ($932) if you reach a management or quality inspection position.”
Autonomous driving data annotation is a low-threshold, repetitive labor. Employees need to draw boxes on the pictures with a mouse and mark the type of car (SUV, sedan, or truck). Each time it is completed, the income is about 2 cents, which is about 0.0031 US dollars. The annotated pictures are divided into 2D and 3D. 2D is like ordinary photos, and 3D is “spatial point cloud data.” A manager of an autonomous driving data annotation company said that with the popularity of lidars, 3D images have become more and more common since 2021.
Although this work is simple, it is very hard. Overtime, staying up late, and even staying up all night are commonplace for autonomous driving data annotation workers. A manager of an autonomous driving data annotation company said that his company does not hire employees over 26 years old.
We noticed that autonomous driving data annotation suppliers are also facing price competition and cost pressures. It is reported that from 2016 to 2018, the unit price of this industry was quite high, and companies that did well could earn a nine-digit income in a year. However, with more teams joining, the choices of customers are increasing, and the industry prices are becoming increasingly transparent. Comparing prices and bargaining with each other are common. The unit price for drawing a box has decreased by 40% to 50% compared to 2016, and there will be a certain degree of decline every year. “Customers are increasingly emphasizing cost reduction and efficiency improvement, and want to draw more boxes with less money.”
In order to reduce costs, some practitioners have also tried to use intelligent tools to improve data annotation efficiency and quality, but this requires a lot of investment and research and development. Before AI completely replaces manual labor, “it is still better to find employees with lower wages to reduce costs.”
