On November 17, 2020, MIT Task Force on the Work of the Future released their final report after their formation in February, 2018 at the request of MIT President Rafael Reif. The final report has 92 pages in length, consisting of three parts in its body: labor markets and growth, technology and innovation, institutional innovation to support workers. As a researcher in computer science, I am especially interested in the part on technology and innovation: not only because I am most familiar with the discussion there, but also because the problems outlined there could be directions that I work on in the future. In this short note, I intend to jot down my understandings after reading that part.

Key abbreviations: AI for artificial intelligence, ML for machine learning.

Insurance. This industry is already digitalized; however, AI and ML has not brought much benefit to it despite many attempts. A related AI technique is called robotic process automation (RPA), which refers to “software that automates rule-based actions performed on a computer”. Until now RPA is not flexible enough to deliver the expected results since the tasks performed by human workers are heterogeneous. Other useful AI techniques are ML-based chatbots for customer service and ML-based evaluation of legal bills. So far the use of AI-based software slows down hiring in relevant departments yet do not lead to large-scale layoffs.

Healthcare. The usages of AI in healthcare are relatively obvious: medical imaging to interpret medical images, natural language processing to read clinical documentation, data science to process massive patients’ data for better predictions. These technologies will enhance the abilities of doctors and nurses, instead of replacing them. However, both the employment share and relative wages of medical transcriptionists declined in the past 20 years - an example of pain for some employees caused by AI.

Autonomous vehicles. I like the description that “autonomous vehicles are essentially high-speed industrial robots on wheels”. Tempting as a driverless future may sound to some, it is unlikely to come by 2030. However, when it does come, it diminishes many jobs including truck drivers, bus drivers, taxi drivers, auto mechanics, insurance adjusters while also creates new ones, both directly and indirectly related.

Warehousing and distribution. Logistics consists of three industries: warehousing and storage, freight trucking, freight trucking arrangements.

  • The warehousing and storage industry is close to being digitalized by warehouse management system, and is undergoing a transformation towards automated gripping systems. Currently, the robots are not dexterous enough for gripping a wide range of things, and the technology is estimated to take three to five years to develop. Then current jobs in picking and packing could be in danger.
  • Local delivery has caused an employment growth in trucking, which may be threatened by mini delivery robots, even if they may only be applicable in well-defined areas. I find the fact interesting that unmanned aircraft in U.S. Air Force require many more people to operate than tranditional aircraft. Likewise, the shift to mini delivery robots will create new jobs, though they require more skills and may not offset the job loss they cause in quantity.
  • Freight trucking arrangement, as a broker connecting firms to truckers, is also facing a change driven by data.

Manufacturing. A figurative scenario is lights-out factories. I am personally a big fan of this concept. I read criticism of the adoption of the industrial internet of things for the first time: the value of the data is unclear. Currently, most small and medium-sized companies in the U.S. do not use robots. For one thing, the robots are simply not dexterous enough right now, and are actually still “early in a long evolution”. For the other, the cost of robots is high: “the price of a robot is only about one-quarter of the total cost”; the programming and integration cost much. With the above being said, current adoption of robots does cause a decrease in employment in some plants.

Additive manufacturing. Allowing for materials like metal, this is a generalized concept of 3D printing, in contrast to the traditional way of subtractive manufacturing. I don’t think the technique itself is related to AI, though AI can help automate complex or customized design, which can then be built via additive manufacturing. Additive manufacturing complements, rather than replace, subtractive manufacturing.