The figure of the robot is very fashionable these days, with books like “The Rise of the Robots” and movies like “Ex Machina” and “Terminator Genisys”. A recent cover of Foreign Affairs was entitled “Hello, robot”
The vision of robots that threaten jobs and perhaps even humanity itself is feeding dire warnings of future problems.; “Invoking the devil” in the evocative expression technology entrepreneur Elon Musk.
However, the current obsession is putting the cart before the mechanized algorithmic oxen, diverting attention from the promises and perils artificial intelligence in the here and now.
The artificial intelligence is now everywhere but basically software. Searches and targeted Google advertising are animated by artificial intelligence, as are movies and products that we recommend Netflix and Amazon, Apple’s Siri digital assistant and Watson set of questions and answers from IBM. Powered by huge amounts of data from all sources, including Web, sensors, smart phones and genomics, the software actually learns his way. The more raw data ingested, the smarter it gets its artificial intelligence.
Meanwhile, the physical world in which there are robots is multidimensional and messy. The same is true of the digital area in which algorithms capable of learning machine, which is called artificial intelligence, has had its best progress in tasks such as face recognition, language translation, forecasting and decision making.
“Many problems are easier on software, so much of the action revolves around artificial intelligence software; logical rather than physical robots, “says Oren Etzioni, computólogo and CEO of the Allen Institute for Artificial Intelligence in Seattle.
In June conducted an interesting test, when more than twenty teams of outstanding engineers robotics gathered in Pomona, California, in a competition sponsored by the Pentagon research agency. The robots had to be guided by simulated environments with hazards, such as a contaminated nuclear plant, and to perform simple tasks, climbing stairs, opening a valve, operate a drill. These tasks would lead to a person about five minutes or ten at the most. The winning robot needed 45 minutes.
Most of the robots battled hard, falling down the steps and taking long pauses to understand things, even using remote control. Turn the knob to open proved a daunting task for many door. A young man in the audience said: “If we care about the Terminator, you just have to leave the doors closed.”
Surely, robots will improve. Google cars that drive alone, for example, are impressive but with heavy rain or snow, it is better to be a human being at the wheel. And robots with humble ambitions and are taking charge of grueling tasks on the floors of factories and helping surgeons to give more precision and control in the operating room.
But the greatest progress has been made on software, also is rapidly coming to the mainstream of the economy. So far, the largest commercial utilization has given the software capable of learning has been in marketing, where it is used to increase the chances of making sales. Marketing tailored, targeted advertising and personalized product recommendations
Both large companies as emerging are beginning to use software capable of learning in decision making at high risk, such as medical diagnosis, prevention of crime, recruitment for hiring and approving loans.
The idea is that any decision can be applied the power of artificial intelligence, to be smarter, fairer and less inclined to the whims and prejudices of men. The goal may be saving money or saving lives.
However, even the most enthusiastic have their doubts.
Take the case of loans for consumption, which is a market where several companies are using big data and algorithms to evaluate the credit risk of the person requesting a loan. It is the digital version of the most fundamental principle of banking: know your customer. The lenders claim that the algorithms can more accurately predict if a candidate is going to pay the debt harvesting data from many sources, from social media connections to how the applicant fills out the online form, that just exam credit history.
The promise subscriptions and prices are more efficient loans, saving consumers billions of dollars. But artificial intelligence applied to loans essentially amounts to a digital black box that looks mountains of data.
“a decision that affects the consumer is taken and he has no idea why it took” says Rajeev Date, former assistant to the Consumer Financial Protection Bureau Director. “That’s disturbing.”
Dr. Herbert Chase, professor of the College of Physicians and Surgeons of Columbia University, participated as an unpaid researcher testing software IBM Watson when scientists that company were adapting it for use in medicine. To try to defeat Watson, Chase recalled a case many years before, in which he made a correct diagnosis of rickets in a young adult woman, but only after numerous tests and months of frustration. He said some symptoms Watson and this, which classifies diagnoses by chance, quickly responded, adult rickets placing second.
It was not perfect, but Dr. Chase was impressed that the technology so much so fast approached. He thinks Watson style software, which you can examine and extract information from thousands of medical articles in seconds, it will be part of medicine in the future, assisting doctors inundated with information and short on time. But concern that worries him is that, over time, “come to rely too much on technology, so it’s equivalent to follow blindly the GPS system of the car to end up in an impasse becomes” he said Chase.
This, of course, is the central problem as intelligent systems evolve. Whether software or physical robots. Will our servants or our masters?
“Although these systems always guess right, follow instructions slavishly robs us of our ability to decide for ourselves,” says Kristian Hammond, artificial intelligence expert at Northwestern University . “Then we would really have a world that nobody wants. The algorithm of the machine would tell us what to do “
The antidote to this is what Hammond calls” transparency “and other designated” narrative “. The explanation of ingredients data are incorporated into a decision Automated and how it is taken. Scientists from universities and corporate researchers are working on artificial intelligence that will give yourself the data audit trails to natural language stories in the chain of reasoning algorithm monitoring technology.
Meanwhile has created new research initiatives that seem to take into account that distant when robots break their chains and become independent from their human masters date. At Stanford, a group of prominent scientists began a study of artificial intelligence designed a hundred years. Musk, meanwhile, has spent millions of dollars in research grants sponsored by the Future Institute of Life, which is dedicated to guide the artificial intelligence beneficial ways.
But at least in the short term, the algorithms do not see much _ _ robots that deserve watchful eyes of man. Some experts warn that what is at stake goes far beyond the field of technology.
“We need to ensure that data and algorithms are continuously subjected to tests and reviews by many different people “says Alex Pentland computational sociologist MIT Media Lab. “Think of representative democracy, creating algorithms in place of law.”
The comments in the section above are dumped responsibility of the participants. Help us to maintain a healthy exchange of ideas, denounced by a “click” on the flag icon, each entry it deems inappropriate. If the comment does not comply with RULES , will be removed. Click here for the rules .
Welcome to our feedback service. We invite you to analyze, comment on and question the articles and reports published by El Diario. To maintain a healthy exchange of ideas to all our users, we need to follow the rules that we detail below. By participating in the feedback service you accept the application of these rules.
You agree that you are fully responsible for the content that you post.
Do not intentionally publish content that violates the copyright, trademark, patent or other intellectual property right of a third person.
No disseminate defamatory, obscene or aggressive that violates the right of a third person to privacy according to municipal, state, federal or even international or is deemed inappropriate. It indemnify Publications Paso del Norte (*), its employees and affiliates of each and every one of the claims and / or damages (including, but not limited to attorneys’ fees) that are submitted by third parties in relation to the content you published. He did not write content that discriminate against others because of their gender, race, ethnicity, nationality, religion, sexual orientation, disability or any other category.
You understand and agree that The Journal is not responsible for content posted by third parties.
it is his understanding that the discussion forums, El Diario published oversees the content and reserves the right to remove, edit or alter content that seems inappropriate for any reason without consent of the author . We reserve the right according to our own discretion to withdraw a user the right to publish content on our site.
You understand and agree that the discussion forums should be used for non-commercial purposes.
request resources or financial support. It will not promote commercial enterprises or perform any commercial activities in our discussion forums.
You understand and agree that the use of any information obtained through the discussion forums at your own risk.
El Diario in no way endorses the broadcast content users, and it can not be made responsible for its accuracy.
With every content that you post, given to The Journal without paying royalties irrevocable, perpetual, exclusive and fully sublicensable license to use, reproduce, modify, adapt, publish, translate, create derivative works from, distribute, perform and display such content in whole or part worldwide and to incorporate them in their work in any way now known or hereafter developed.
No comments:
Post a Comment