Ctrl + N
Civil rights advocates are raising concerns about the Trump administration's idea to roll back anti-discimination rules -- insisting these decades-old provisions that prohibit discrimination in employment and housing are also an effective way to combat bias in artificial intelligence.
NAACP Legal Defense Fund President Sherrilyn Ifill slammed the Trump administration after a leaked internal Justice Department memo revealed officials were considering ways to scale back or remove so-called “disparate impact requirements." Ifill noted that these rules saying actions that disproportionately impact protected groups can be considered discriminatory -- even if that was not the intent -- take on even more importance in the AI era.
“This is precisely the moment in which the Trump administration is trying to get rid of disparate impact, when in fact disparate impact is one of the most important tools we have in the employment context and housing and other areas,” she said at a Washington Post Live Event. “But particularly, it has the power to be incredibly useful in the AI field, where essentially what you need to do is figure out how is this instrument you're using working, what is the result it's producing and then be able to work our way from there to understand if there's bias in the instrument.”
Policymakers are increasingly worried about the potential for algorithms to unfairly discriminate against people. Ifill's comments put a fine point on a dilemma for lawmakers — who are trying to understand where laws already on the books could be applied to combat bias in artificial intelligence, and where they might fall short and need new safeguards. Any significant change to civil rights law could make the problem even more complex.
Technologists say bias is often a result of the data sets the tools are trained on. Last year, for instance, Amazon had to shut down an AI recruiting tool because it was discriminating against women. The company realized the system was doing this because it had been trained on a decade of résumés submitted to the company — which were disproportionately from men, as is often the case in the technology industry. (Amazon founder and chief executive Jeffrey P. Bezos owns The Post.)
But many in industry are skeptical that Washington will be up to the challenge. After seeing lawmakers struggle to grasp basic tech concepts during hearings with executives like Mark Zuckerberg, Bari Williams, an executive at Bay Area start-up All Turtles, said at CES in Las Vegas that she doesn’t know if they’ll be able to take on the complex task of regulating bias in algorithms.
“You cannot set policy and regulation if you don’t understand how it works,” Williams said.
Brian Schatz (D-Hawaii) has proposed legislation to give more authority, staff and resources to the Federal Trade Commission, which could be one way to ensure the federal government can extend more oversight of new ways technology can harm consumers, including bias in artificial intelligence. The provisions were included in a privacy bill he introduced last year, which has gained broad support among Senate Democrats, but it has no Republican co-sponsors.
Sunmin Kim, who is a technology adviser to Schatz, told CES attendees on Thursday that policymakers need to look at whether bias is playing out in new tools, such as price targeting. E-commerce companies are increasingly trying to get customers to pay as close to their maximum price as possible by offering different customers different prices based on their geographic location, or age. Although this technology might be good for commerce, Kim said, more oversight is needed to see whether companies are determining prices on the basis of race or other protected categories.
In addition to more oversight, Kim thinks the government can help address bias in the training data sets that shape the algorithms. Last month, Congress passed the Open Government Data Act, which aims to make every government data asset available to the public in a machine-readable format. Kim said these data sets may provide less biased data for algorithms to train on. The bill is awaiting President Trump’s signature.
As lawmakers grapple with the steps they can take to rein in bias in artificial intelligence, panelists at CES made it clear the onus also needs to be on the companies to be more transparent.
Kim said her office has started to push the companies to tell them more about their algorithms — something technology businesses often resist because the algorithms are important intellectual property.
“As they make more and more decisions,” she said, “it’s not unreasonable for the government to ask, ‘What’s inside the black box?’ ”
|You are reading The Technology 202, our guide to the intersection of technology and politics.|
|Not a regular subscriber?|
BITS: James Martin, a Google shareholder, sued the tech giant's board of directors and senior managers for allegedly covering up sexual misconduct, The Washington Post's Elizabeth Dwoskin reported. The suit “stems from revelations last year that the company awarded a $90 million exit package to executive Andy Rubin, the creator of the Android operating system, even as Rubin was being pushed out over sexual misconduct accusations,” according to my colleague.
Martin claimed that Google's leaders breached their duty to the company's investors because of the alleged coverup. The suit's targets include Larry Page and Sergey Brin, co-founders of the company, executive chairman Eric Schmidt, chief executive Sundar Pichai and others. “The suit charges prominent venture capitalists John Doerr and Ram Shriram, who led the board’s leadership development and compensation committee, with approving the payment while being fully apprised of the allegations,” Elizabeth wrote.
NIBBLES: Apple wants to strengthen its services business — which include App Store sales, mobile payments and subscriptions to music streaming — in the face of waning demand for its iPhones, the Wall Street Journal's Tripp Mickle reported. Although the company's Maps app and Apple Pay initially faced a rough start, iTunes and the App Store have been successful. Still, the company faces several hurdles in its effort to grow its services. Sales from the App Store have slowed down and companies such as Netflix are seeking to circumvent Apple's payment service.
A legal case could also further the company's problems. “Meanwhile, Apple awaits a Supreme Court decision in a case that determines if iPhone owners can sue Apple for antitrust damages for its alleged monopoly on the App Store,” Mickle wrote. “Should the justices decide the case has merit, Apple would face a challenge to its 30% cut of App Store sales.”
BYTES: A survey found that more Americans support artificial intelligence than oppose it, according to the Verge's James Vincent. “In polling organized by the University of Oxford’s Future of Humanity Institute, forty-one percent of respondents said they somewhat or strongly supported the development of AI, while 22 percent said they somewhat or strongly opposed it,” Vincent wrote. “The remaining 28 percent said they had no strong feelings one way or the other.”
A vast majority of respondents supported regulation of AI, with 82 percent saying they somewhat or strongly agreed that “robots and artificial intelligence are technologies that require careful management,” the Verge noted. The firm YouGov conducted the study for the Center for the Governance of AI, part of the Future of Humanity Institute. As Bloomberg News's Jeremy Kahn reported, the survey also found that more than two-thirds of respondents said they either had “no confidence” or “not too much confidence” in Facebook to develop AI.
— Chinese telecommunications giant Huawei's troubles are deepening. Polish authorities charged a Chinese Huawei executive for carrying out espionage for China, the Wall Street Journal’s Drew Hinshaw and Dan Strumpf reported. A Polish citizen who was deputy head of the Polish Internal Security Agency’s IT security department was also detained.
“Officers of Poland’s counterintelligence agency this week searched the local Huawei office, leaving with documents and electronic data, as well as the home of the Chinese national, Poland’s state-owned broadcaster reported Friday,” according to the Journal. “The individual wasn’t named, but was identified as a graduate of one of China’s top intelligence schools, as well as a former employee of the Chinese consulate in the port city of Gdansk.”
The arrest comes after the company's chief financial officer was detained in Canada in December at request of the U.S. on charges related to Iranian sanctions. "Unlike those allegations, the nature of the charges in Poland speak directly to suspicions by Washington and other Western governments that Huawei could be used by Beijing as a global spying tool," the Wall Street Journal reported.
— More tech news from the private sector:
As the Trump administration has said it plans to tighten the H-1B visa program allowing foreigners to work in “specialty occupations,” more foreign workers in Silicon Valley are considering moving to Canada to continue their careers, according to my colleague Emily Rauhala. One Canadian entrepreneur, Irfhan Rawji, founded the company MobSquad to assist tech firms in moving skilled workers such as software engineers to Canada and often visits Silicon Valley to pitch his services to U.S. executives.
“Though it is hard to track how many foreign nationals have moved from the United States — the Canadian government tracks newcomers by country of citizenship, not residence — immigration lawyers and recruiters on both sides of the border say the number of inquiries from nervous H-1B holders has skyrocketed since 2017,” my colleague wrote.
— Sen. Doug Jones (D-Ala.) formally asked the Federal Election Commission to investigate online disinformation efforts to influence a U.S. Senate race in 2017 in Alabama — which he won, the New York Times's Alan Blinder reported.the New York Times's Alan Blinder reported. The Post previously reported that Jones had no knowledge of the effort and was calling for a federal probe.
“Such deceptive tactics have no place in American politics and must be repudiated by those involved in our political system,” Jones said in a letter to Ellen L. Weintraub, a Democratic member of the FEC, according to the Times.
— More technology news from the public sector:
— Callisto, a nonprofit organization based in San Francisco, is developing software to help report sexual assault in the workplace, Jess McHugh reported in the Wall Street Journal. Callisto, which has also partnered with colleges to help report assaults with Callisto Campus, solicited more than 1,600 founders of start-ups to begin testing the version of its service for the workplace in November of last year.
The software, called Callisto Expansion, asks victims to enter information about the assault and the assaulter. “The information is then encrypted and stored in a secure database that not even Callisto can access,” according to the Journal. “If two users describe the same perpetrator — even years apart — they are contacted by separate Callisto legal counselors (working pro bono or at a discount), who provide information about reporting options, including filing a complaint with human resources or the police.” Callisto plans to expand its service this year in several industries.
— Experts said large groups of people around the world will probably remain without access to Internet for decades barring major improvements in education and broadband infrastructure, the Guardian's Ian Sample reported. “While half the world’s population now uses the internet, a desperate lack of skills and stagnant investment mean the UN’s goal of universal access, defined as 90% of people being online, may not be reached until 2050 or later, they said,” according to the Guardian.
— More tech news generating buzz around the Web:
— Ring, a company making home security cameras that was acquired by Amazon, adopted loose security practices in its management of customers' video feeds, potentially allowing employees to watch people inside their homes, according to the Intercept's Sam Biddle. “Beginning in 2016, according to one source, Ring provided its Ukraine-based research and development team virtually unfettered access to a folder on Amazon’s S3 cloud storage service that contained every video created by every Ring camera around the world,” the Intercept reported. “This would amount to an enormous list of highly sensitive files that could be easily browsed and viewed.”
Meanwhile, the company also allowed executives and engineers in the United States access to live feeds from some customers' cameras even if they did not need it in their jobs. Those with such access just needed the email address of a customer to be able to watch their camera feeds, but Biddle wrote that a “source said they never personally witnessed any egregious abuses.” (Amazon founder and chief executive Jeffrey P. Bezos owns The Washington Post.)
After the story's initial publication, “Ring spokesperson Yassi Shahmiri told The Intercept that ‘Ring employees never have and never did provide employees with access to livestreams of their Ring devices,’ a claim contradicted by multiple sources,” Biddle wrote.
This item has been updated to reflect Ring's response to the Intercept story.
— More news about tech incidents and blunders:
— Today in funding news:
- CES technology show in Las Vegas.
- The Brookings Institution hosts a discussion titled “How China and the U.S. are advancing artificial intelligence” on Jan. 14.
What's Deepak Chopra doing at CES of all places?
Trump says he didn't think Mexico would pay for the wall with “a check”:
Is your food safe to eat? Shutdown forces FDA to sharply reduce food inspections: