Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy by Cathy O'Neil Page A
Authors: Cathy O'Neil
Tags: General, Social Science, Political Science, Business & Economics, Public Policy, Statistics, Privacy & Surveillance
Ads: Link
codes, with special attention to those who have clicked on an ad for payday loans or seem to be concerned with post-traumatic stress. (Combat veterans are highly recruited, in part because it’s easier to get financing for them.)
The campaign proceeds to run an endless series of competing ads against each other to see which ones bring in the most prospects. This method, based on so-called A/B testing, is one that direct-mail marketers have been using for decades. They send a plethora of come-ons, measure the responses, and fine-tune their campaigns. Every time you discover another credit card offer in your mailbox, you’re participating in one of these tests. By throwing out the letter unopened, you’re providing the company with a valuable piece of data: that campaign didn’t work for you. Next time they’ll try a slightly different approach. It may seem fruitless, since so many of these offers wind up in the trash. But for many direct marketers, whether they’re operating on the Internet or through the mail, a 1 percent response rate is the stuff of dreams. After all, they’re working with huge numbers. One percent of the US population is more than three million people.
Once these campaigns move online, the learning accelerates. The Internet provides advertisers with the greatest laboratory ever for consumer research and lead generation. Feedback from each promotion arrives within seconds—a lot faster than the mail. Within hours (instead of months), each campaign can zero in on the most effective messages and come closer to reaching the glittering promise of all advertising: to reach a prospect at the right time, and with precisely the best message to trigger a decision, and thus succeed in hauling in another paying customer. This fine-tuning never stops.
And increasingly, the data-crunching machines are sifting through our data on their own, searching for our habits and hopes, fears and desires. With machine learning, a fast-growing domain of artificial intelligence, the computer dives into the data, following only basic instructions. The algorithm finds patterns on its own, and then, through time, connects them with outcomes. In a sense, it learns.
Compared to the human brain, machine learning isn’t especially efficient. A child places her finger on the stove, feels pain,and masters for the rest of her life the correlation between the hot metal and her throbbing hand. And she also picks up the word for it: burn. A machine learning program, by contrast, will often require millions or billions of data points to create its statistical models of cause and effect. But for the first time in history, those petabytes of data are now readily available, along with powerful computers to process them. And for many jobs, machine learning proves to be more flexible and nuanced than the traditional programs governed by rules.
Language scientists, for example, spent decades, from the 1960s to the early years of this century, trying to teach computers how to read. During most of this time, they programmed definitions and grammatical rules into the code. But as any foreign-language student discovers all too quickly, languages teem with exceptions. They have slang and sarcasm. The meaning of certain words changes with time and geography. The complexity of language is a programmer’s nightmare. Ultimately, coding it is hopeless.
But with the Internet, people across the earth have produced quadrillions of words about our lives and work, our shopping, and our friendships. By doing this, we have unwittingly built the greatest-ever training corpus for natural-language machines. As we turned from paper to e-mail and social networks, machines could study our words, compare them to others, and gather something about their context. The progress has been fast and dramatic. As late as 2011, Apple underwhelmed most of techdom with its natural-language “personal assistant,” Siri. The technology was conversant only in certain areas,
The campaign proceeds to run an endless series of competing ads against each other to see which ones bring in the most prospects. This method, based on so-called A/B testing, is one that direct-mail marketers have been using for decades. They send a plethora of come-ons, measure the responses, and fine-tune their campaigns. Every time you discover another credit card offer in your mailbox, you’re participating in one of these tests. By throwing out the letter unopened, you’re providing the company with a valuable piece of data: that campaign didn’t work for you. Next time they’ll try a slightly different approach. It may seem fruitless, since so many of these offers wind up in the trash. But for many direct marketers, whether they’re operating on the Internet or through the mail, a 1 percent response rate is the stuff of dreams. After all, they’re working with huge numbers. One percent of the US population is more than three million people.
Once these campaigns move online, the learning accelerates. The Internet provides advertisers with the greatest laboratory ever for consumer research and lead generation. Feedback from each promotion arrives within seconds—a lot faster than the mail. Within hours (instead of months), each campaign can zero in on the most effective messages and come closer to reaching the glittering promise of all advertising: to reach a prospect at the right time, and with precisely the best message to trigger a decision, and thus succeed in hauling in another paying customer. This fine-tuning never stops.
And increasingly, the data-crunching machines are sifting through our data on their own, searching for our habits and hopes, fears and desires. With machine learning, a fast-growing domain of artificial intelligence, the computer dives into the data, following only basic instructions. The algorithm finds patterns on its own, and then, through time, connects them with outcomes. In a sense, it learns.
Compared to the human brain, machine learning isn’t especially efficient. A child places her finger on the stove, feels pain,and masters for the rest of her life the correlation between the hot metal and her throbbing hand. And she also picks up the word for it: burn. A machine learning program, by contrast, will often require millions or billions of data points to create its statistical models of cause and effect. But for the first time in history, those petabytes of data are now readily available, along with powerful computers to process them. And for many jobs, machine learning proves to be more flexible and nuanced than the traditional programs governed by rules.
Language scientists, for example, spent decades, from the 1960s to the early years of this century, trying to teach computers how to read. During most of this time, they programmed definitions and grammatical rules into the code. But as any foreign-language student discovers all too quickly, languages teem with exceptions. They have slang and sarcasm. The meaning of certain words changes with time and geography. The complexity of language is a programmer’s nightmare. Ultimately, coding it is hopeless.
But with the Internet, people across the earth have produced quadrillions of words about our lives and work, our shopping, and our friendships. By doing this, we have unwittingly built the greatest-ever training corpus for natural-language machines. As we turned from paper to e-mail and social networks, machines could study our words, compare them to others, and gather something about their context. The progress has been fast and dramatic. As late as 2011, Apple underwhelmed most of techdom with its natural-language “personal assistant,” Siri. The technology was conversant only in certain areas,
Similar Books
