The discussions around automation and artificial intelligence appears to be ruled by either doomsayers who are in a fear that robots will replace all humans in the labor force, or optimists who believe that there is nothing new below the sun. But Erik Brynjolfsson, the MIT Sloan professor, and his associates claim that this debate requires to take a diverse tone.
New study discovers that particular tasks within jobs, instead of whole occupations themselves, will be substituted by computerization in the years to come, with some roles more greatly affected than others.
“Our discovery suggests that a change is required in the argument about the impacts of AI. These changes are away from the ordinary aim on complete computerization of full jobs and an omnipresent occupational substitution toward the reengineering of business processes and redesigning of jobs,” the scientists claimed in an article posted in the American Economic Association Papers and Proceedings in May. The article is written by Brynjolfsson, professor of machine learning department at Tom Mitchell of Carnegie Mellon University. He was assisted by Daniel Rock, a researcher at the MIT Initiative and doctoral candidate on the Digital Economy.
“In spite of what Hollywood is stating, we are extremely far from general artificial intelligence. That is, AI that can just do anything a human can do,” Brynjolfsson claimed further. “We do not have anything near to that. We will not have that for decades, unless there is some amazing penetration.”
On a similar note, a supplier of Enterprise AI solutions, Indico, this week declared the roll out of a new open source program aimed on making easy the employment of transfer learning by means of natural language. Being an open-source library, Enso is developed to transfer learning methods and streamline the benchmarking of embedding for a huge range of natural language processing jobs.