News reporter trainee reporter Han Han chess intern Sun Leying
As time went back to 2000, Goldman Sachs employed 600 tellers at the U.S. stock exchange in New York. Today, Goldman Sachs only has 2 stock traders. Working with them is an automated trading program. As artificial intelligence has become more and more capable in all walks of life, many people are beginning to worry that their jobs will be taken away by them.
In 2013, a study by Oxford University predicted that by 2034, 47% of occupations will be automated. The study surveyed more than 700 occupations and listed long “death lists†based on the probability of automation. Among them, the probability that AI’s robberies for jobs such as drivers, salesmen, and most financial industry tellers is as high as 98%.
In addition to these highly repetitive and partial-mechanical tasks, AI's breakthroughs in semantic analysis, sentiment recognition, and dialogue interactions have also led many to suspect that jobs such as lawyers that focus on "human nature" are also falling. David H.Autor, professor of economics at the Massachusetts Institute of Technology, believes that the U.S. economy is being "hollowed out." In the U.S. economic pyramid, the growth of upper-level jobs has slowed down due to automation. Middle-level jobs are being automated or outsourced, and only the bottom is still adding new jobs. Therefore, he concluded that although artificial intelligence did not cause unemployment, its development provided bottom-level jobs.
However, opponents of the "AI threat theory" believe that the speed of professional automation is greatly overestimated. The New York Times recently published an article titled "I'm sorry, your lawyer won't be replaced with a robot." It introduced the study of Dana Remus, professor of law at the University of North Carolina, and Frank Lev, professor of economics at the Massachusetts Institute of Technology. Conclusion: At least in the legal field, automation is still very slow.
Detective General Computer: Combine Relationship Networks to Find Suspects
In 2011, artificial intelligence entered the legal arena. The New York Times published a review article entitled "Cheap Software Is the Enemy of Expensive Lawyers," highlighting the surprising efficiency of software in dealing with legal documents. For example, Blackstone Discovery, headquartered in Silicon Valley, is committed to developing lawyer robots, quoting $100,000, and they can help analyze 1.5 million legal documents.
This kind of analysis is not only a traditional keyword search. With more than 500,000 corporate mails after the collapse of the Enron Corporation in the US, it has been used as an artificial intelligence research. Computers have both semantic analysis and social analysis of texts. breakthrough.
In terms of semantic analysis, in addition to the software's direct search for keywords, there is a set of keyword-related "filters" that capture more indirect relevance. For example, by using the keyword “dogâ€, the computer will also help you find content about “the best friend of mankind†or “embarrassmentâ€.
The sociological analysis can better reflect the general reasoning ability of computer detectives. The system records the interactions of emails, text messages, and telephones. It can sort out an entire social chain and capture the anomalies in these social chains. For example, the switch point between social media is an abnormal point in this chain. If there is a "call me" message in the text message or mail, this means that there may be employees who want to do something private.
In addition, computers also have the ability to recognize emotions. It can identify positive and negative emotions, and can also read and emphasize the tone of emphasis. The sudden fluctuation in the style of these writings is also an abnormal point in the chain of reconnaissance. Some writing style changes are even beyond human awareness. For example, if you are afraid of your email being monitored by the FBI, your use of infinitives will decrease.
With these technical capabilities, computer software not only quickly finds content related to the case in a large number of documents, but also helps humans comb out the entire relationship network, visualizes the company’s internal operations, and finds suspicious private information. transaction.
Not the opponent is the helper
However, does artificial intelligence steal the lawyer's job? Dana Remus, professor of law at the University of North Carolina, and Frank Lev, professor of economics at the Massachusetts Institute of Technology, believe that discussing how many percent of careers will be automated is all about paperwork. Only in-depth analysis of the specific conditions of each industry can be reliable. in conclusion.
They combined data from Consilio's Sky Analytics and divided the work of lawyers into 13 categories to analyze the role of computers in them. The results of the study are shown in the table below:
It can be seen that only 4.1% of the work tasks in Tier 1 companies are easily replaced by computers, and this ratio is only 3.6% in two to five companies. This low proportion is mainly due to two aspects. First, "document review" only belongs to "factual verification" in these categories, rather than due diligence (which is not easy to implement). Second, because some clients explicitly request lower litigation fees, many large companies do not appear on regular tasks (such as document review) but instead hand over to some outsourcing companies. As a result, customers have been exerting pressure on law firms over the past ten years in an attempt to allow companies to control costs through outsourcing.
In summary, artificial intelligence on the one hand bears many “low-level†jobs that are often outsourced by the law, while on the one hand reducing lawyers’ “advanced†workload. The paper also pointed out that if all the technology in the legal field is put into use, lawyers will save about 13% of the time. Taking into account the unemployment rate, the more realistic data is that the workload of lawyers will decrease by 2.5% per year within five years. Therefore, instead of artificial intelligence, it would be better to take away the lawyers' jobs than to say that artificial intelligence will be a good assistant for lawyers.
“There is a common misunderstanding, as if one of the professions is being automated, and the rest is also something sooner or later,†Levy said. “This situation is actually very rare.â€
This research may provide some inspiration for automation research in other professions. When discussing the pros and cons of robots or artificial intelligence, it may be useful to make a more detailed division of work and take into consideration factors outside the machine, such as market adaptability and existing industry rules.
This research conclusion can be confirmed by the development status of some lawyer robots, such as Ross Ross, a startup company. Ross can immediately find a legal case on the online legal database that is closest to the case, and a human lawyer spends 10 hours. However, it took humans 2 and a half years to develop Ross's powerful search capabilities.
In addition, Ross is also updating his ability to write memos. Enter a legal question and Ross will respond with a few general answers and a 2-page explanation memo one day later. The reason why it takes one day is because Ross's “writing style†is still very bad, and people need to process the drafts “written†by mankind.
From this point of view, artificial intelligence is still far away from taking away the lawyer's job. Even if it is to grab, it is gradually robbing one step at a time.
Usb-C Hub 4 In 1,Usb C Hub Laptop Docking Station,Usb C Hub For Surface Book Wireless,Docking station for laptop,Docking station for macbook pro
Shenzhen Konchang Electronic Technology Co.,Ltd , https://www.konchangs.com