[ad_1]
Rep. Alexandria Ocasio-Cortez criticized Amazon for what she called "bias" in the technology giant's face detection technology.
The first-year congressman told his 2.7 million followers of Twitter: "When you do not talk about human prejudices, this bias is automated.The machines are a reflection of their creators, which means that they are flawed, and we should be aware of it.It 's one of the good reasons why diversity is not just "beautiful", that' s it. is a protection against trends like this. "
The Ocasio-Cortez study mentioned in its tweet, which was conducted by researchers at MIT and the University of Toronto, revealed that Amazon's technology qualified women to darker skin of men 31% of the time. Women with lighter skin were poorly identified 7% of the time.
APPLE SCRAMPS TO FIX FACETIME BUG THAT ALLOWS PEOPLE TO HOPE YOU
Matt Wood, Director of Artificial Intelligence at the cloud computing unit D & # 39; Amazon, told the Associated Press that the study uses a technology of "facial analysis" and not "facial recognition". According to Wood, facial analysis "can spot faces in videos or images and assign generic attributes such as wearing glasses; recognition is a different technique of associating a face with faces in videos. and pictures. "
Wood also said that Amazon had updated its technology since the study – which conducted testing in August of last year – and its own analysis showed "zero false positive result" [19659003] However, the researcher at MIT's laboratory, Joy Buolamwini, responded in an article on Medium that the public "can not expect" Amazon to control itself or provide untested and unregulated technology to the police or to government agencies. "
" The terminology used in the field is not always consistent, and terms such as "facial recognition" or "facial recognition" are used interchangeably "Buolamwini, who is pursuing a Ph.D. Participatory AI at the Center for Civic Media at MIT Media Lab, wrote on Medium. "Often, companies such as Amazon provide artificial intelligence services analyzing faces in different ways, such as Indication of sex or provision of identification services. All these systems, whatever their name, must be the object of a permanent search to detect any prejudicial prejudice. "[19659003] FACEBOOK CLICKES ON THE REPORT OF THE DIRECTOR GENERAL OF OLD HARVARD, THE CEO MARK ZUCKERBERG , claims half of its users.
Amazon has been criticized by lawmakers, human rights groups and shareholders for its facial recognition technology called "Rekognition". "
A group of shareholders said investors feared that facial recognition technology would be used by local and federal government agencies "to justify surveillance, detention of people seeking to enter the United States "and she urges Amazon to collaborate with experts on human rights and civil liberties to assess the impact of the program.
A 2018 ACLU study, contested by Amazon, showed that the company's facial recognition technology incorrectly identified 28 members of Congress as police suspects, fueling concerns about racial prejudice.
In his public blog, Amazon states that facial recognition technology should not replace human judgment.
"In all public safety and law enforcement scenarios, technologies such as Amazon Rekognition should only be used to restrict Amazon's Rekognition responses allow managers to quickly obtain a set of faces that can be subjected to in-depth human analysis Given the severity of safe use cases, human judgment is needed to improve facial recognition, and facial recognition software should not to be used independently, "says the company.
Amazon also notes that facial recognition technology, which is already in use US airports help to secure travelers more quickly and can also be used to identify victims of human trafficking and to prevent fraud when customers use certain financial services applications.
CHINA DEVELOPING AN APP TO MONITOR "DEADBREAKERS" ]
Technologists, activists and its own employees are likely to keep up the pressure on Amazon. The algorithms, which range from Siri to the results displayed in Google's search engine, can be misused and convict innocent people in jail.
"Algorithms are always created by human beings, and these algorithms are still linked to fundamental principles. Ocasio-Cortez, who recently said that "technology monopolies" were one of the biggest threats to journalism, told the writer Ta-Nehisi Coates at the annual MLK Now event . "These are just automated assumptions. And if you do not correct the bias, you simply automate the bias. "
Data scientist Emily Gorcenski defended Ocasio-Cortez's comments on skewed algorithms.
" This phenomenon is well recognized in the field ", Gorcenski wrote on Twitter . "Not only is the phenomenon well understood, but its fundamental behaviors are essential to being a data specialist. All data scientists recognize that algorithms will saturate biases without controls. There is a whole subfield devoted to studying ways to avoid this. "
The Associated Press contributed to this report.
[ad_2]
Source link