This raises an unsettling question - when will a machine take your job? There’s no definitive answer yet, but some of the world’s leading artificial intelligence researchers hope to find out. Katja Grace, a research associate at the Future of Humanity Institute at the University of Oxford, and colleagues from the AI Impacts project and the Machine Intelligence Research Institute surveyed 352 scientists and used their answers to predict how long it would take for machines to surpass humans in various tasks. They contacted many of the world's top machine learning experts, including Yann LeCun, director of AI research at Facebook, Mustafa Suleyman of Google DeepMind, and Zoubin Ghahramani of Uber's AI Lab. The good news is that many people's jobs are likely to be safe for some time to come. Researchers estimate that there is a 50% chance that machines will replace everyone's job within the next 120 years. “The most surprising thing is that the predictions are so late,” Grace said. “I would have expected that given the rapid progress in machine learning in recent years and the fact that our respondents were all machine learning researchers, the predictions would have been earlier.” So what exactly does this mean for the coming years and decades? Increased unemployment?This survey suggests that machines will be able to fold your laundry by 2021. So if you work in a laundromat, is it time to throw in the towel? I'm afraid not. Machines that can fold clothes already exist: Roboticists at the University of California, Berkeley, have developed a robot that can expertly fold towels, jeans, and T-shirts. Granted, it took a robot about 19 minutes to pick up, check and fold a piece of clothing in 2010, but by 2012 it took just over six minutes to fold a pair of jeans and a T-shirt. But perhaps most surprising is that robots can do the tedious task of pairing socks. However, despite this progress, it will still be some time before such robots can truly replace humans. "I am skeptical about the timeline for certain tasks that require hands-on work to be replaced by machines," said Jeremy Wyatt, professor of robotics and artificial intelligence at the University of Birmingham. “It’s one thing to do it in a lab, but it’s another to get a robot to do it better than a human in the real world.” Manipulating objects in the real world is an incredibly complex task for machines, requiring both an understanding of what to manipulate and how to manipulate it in a randomly changing environment. Tasks that don’t require actual manipulation are easier to master. Wyatt believes that robotic mobility — including driverless cars and automated deliveries — is about what the internet was in the early 1990s. “Moving things around is probably another 10 years away.” Robot assistantTowel folders are safe now, but truck drivers and retail clerks do have reason to think about their careers in the next 20 years. Researchers predict that artificial intelligence will be able to drive trucks by 2027 and perform retail work by 2031. The traditional retail assistant job — the friendly clerk who helps you find a pair of jeans in the store and shows you how they fit — requires complex physical and communication skills. For now, the job is probably safe. But Wyatt said that as more people shop online, artificial intelligence in the form of chatbots and algorithms may be more easily able to replace other jobs in the retail industry than we think. "If you look at how many transactions we currently conduct online are mainly automated programs - a large part of them. They are already using a certain amount of artificial intelligence." Humans are afraid Perhaps the most difficult tasks for machines to master are those that even humans need years of training to master. These often involve intuitive decision-making, complex physical environments, or abstract thinking—all of which computers have difficulty with. Experts predict that robots will not replace surgeons until around 2053, and it may take 43 years before they can compete with mathematicians in top academic journals. They also predict that by 2049, novels written by AI could be on the New York Times bestseller list. In fact, machines have already begun to make inroads into this field. Google has been training its AI programs to write romance novels and news reports in the hope that they will be more creative. An AI robot named Benjamin can also write short science fiction movie scripts - even if some of the content makes no sense at all. Then there is the work of Automated Insights, whose algorithms have generated millions of personalized news, finance and sports articles for Reuters and the Associated Press. Adam Smith, chief operating officer of Automated Insights, said the technology is meant to complement human work, not replace it. “Automated journalism creates content that didn’t exist before, but humans are still needed to add context to those stories.” These reports are produced according to established patterns, after extracting information from huge data sets and adding it to the template. It will still take 30 years to create a best-selling novel with beautiful words and attractive plots. The end result of asking machines to create language is often just a poor imitation. Wyatt said the challenge now is getting AI to produce material that humans can accept. “Anything slightly below human level is unacceptable to us,” he said. “Take chatbots for example – they are not far off human performance… but we are very sensitive to the ridiculous errors contained in any of their flaws.” Grace believes that this survey can remind people that the world is about to undergo a huge change: "Technically speaking, I don't think there is any task that humans can do that artificial intelligence can't do." But she believes some jobs may never be replaced by machines. For example, as long as churchgoers want a human to be the pastor, that job may never be replaced by a robot. “There are still some jobs that can only be done by people because we care about who they are,” she said. access This article is from BBC |
<<: YouTube to launch monthly ad-free membership, but TV stations not support it
>>: The VR Fund: AR company industry map in the second quarter of 2017
David Arnott , editor of the Nashville Business J...
The user fission of Love Bank is very ingenious. ...
What changes will the quality of server rental br...
Beijing Daily reported that two days ago, residen...
Loading long image... Source: National Geographic...
Author: Han Bing, Chief Physician, Peking Union M...
Apple released its next-generation operating syst...
Everyone has different habits when it comes to cl...
Popular Science Times reporter Chen Jie Shi In th...
Novel motion lighting control technology In today...
Generally speaking, people are more concerned abo...
Recently, staff from the Xishuangbanna National N...
Official Weibo operation plan: 1. Planning Purpos...
This is a message from the third year of Shenjue ...
There is a saying circulating on the Internet: Do...