model

NYU’s crowdsourced questions probe extent of language model bias

In a new study, researchers at New York University (NYU) found that popular language models including Google’s BERT and ALBERT and Facebook’s RoBERTa reinforce harmful race, gender, socioeconomic, religious, age, sexual, and other stereotypes. While previous research has uncovered bias in many of the same models, this latest work suggests the biases are broader in scope than originally thought.

Pretrained language models like BERT and RoBERTa have achieved success across many natural language tasks. However, there’s evidence that these models amplify the biases present in the data sets they’re trained on, implicitly perpetuating harm with biased representations. AI researchers from MIT, Intel, and the Canadian initiative CIFAR have found high levels of bias from BERT, XLNet, OpenAI’s GPT-2, and RoBERTa. And researchers at the Allen Institute for AI claim that no current machine learning technique sufficiently protects against toxic outputs, highlighting the need for better training sets and model architectures.

Read More Read more
women

UK passport photo checker shows bias against dark-skinned women

An illustration showing photos of three people with different skin tones. The photo of the darkest skinned person has a poor quality score and the photo of the lightest skinned person has a good quality score
An illustration showing photos of three people with different skin tones. The photo of the darkest skinned person has a poor quality score and the photo of the lightest skinned person has a good quality score

Women with darker skin are more than twice as likely to be told their photos fail UK passport rules when they submit them online than lighter-skinned men, according to a BBC investigation.

One black student said she was wrongly told her mouth looked open each time she uploaded five different photos to the government website.

This shows how “systemic racism” can spread, Elaine Owusu said.

The Home Office said the tool helped users get their passports more quickly.

“The indicative check [helps] our customers to submit a photo that is right the first time,” said a spokeswoman.

“Over nine million people have used this service and our systems are improving.

“We will continue to

Read More Read more
women

Women Are Twice As Likely To Experience “Every Form Of Bias”

A survey conducted by the Hollywood Commission on Eliminating Sexual Harassment and Advancing Equality has found that women throughout the entertainment industry and across every demographic are faced with “every form” of bias and discrimination as they navigate their careers.

The commission, chaired by Anita Hill, released its survey concerning bias in the entertainment industry three years after its formation, as the #metoo and #timesup movements raised awareness and in Hollywood. While the trends it has uncovered aren’t that surprising to anyone familiar with the industry, the breadth and depth of the problem as revealed by the survey data is shocking.

The report opens with a stark summary:

“Those who are underrepresented across the entertainment industry (women/non-binary, people of color, LGBTQ identifying, and individuals with a disability) continue to face significant bias. On average, underrepresented groups reported experiencing two to three times as much bias as majority men.”

Concretely, less

Read More Read more
women

Women outwit Hollywood bias with help from industry insiders

LOS ANGELES (AP) — Kaitlyn Yang knows it’s rare for women to work in visual effects but wanted to find out just how much company she has.



In this image provided by the Television Academy, Layne Eskridge, creative executive at Apple TV, center, takes part in "Meet Our Alumni: Intern to Industry Professional," one of the Television Academy Foundation's intern Speaker Series events on Aug. 2, 2018, in Los Angeles. Women of color who work behind the camera in fields including writing and visual effects are finding career support, including from the Television Academy Foundation and its internships. But four former interns say the industry must do more to foster diversity. (Photo by Jordan Strauss/Invision for the Television Academy/AP Images via AP)


© Provided by Associated Press
In this image provided by the Television Academy, Layne Eskridge, creative executive at Apple TV, center, takes part in “Meet Our Alumni: Intern to Industry Professional,” one of the Television Academy Foundation’s intern Speaker Series events on Aug. 2, 2018, in Los Angeles. Women of color who work behind the camera in fields including writing and visual effects are finding career support, including from the Television Academy Foundation and its internships. But four former interns say the industry must do more to foster diversity. (Photo by Jordan Strauss/Invision for the Television Academy/AP Images via AP)

Devising an informal survey earlier this year, she painstakingly searched 24,000 LinkedIn entries for female visual effects supervisors in North America. Her tally:

Read More Read more
style

Writer pens a $5M seed round for its AI style guide that flags bias and tone

Anyone who writes online or in a word processor has likely gotten used to the inevitable squiggly line denoting a misspelled word or clumsy phrase. But what if you use a word that’s loaded, a phrase that’s too formal or not formal enough, or refer to a group of people in an outdated way? Writer is a service that watches as you type, flagging language that doesn’t match up with your style guide and values, and it just raised $5 million to scale up.

Both people and the companies they work for want to improve the way they write, but not just in terms of grammar and spelling. If a company says it’s inclusive, but the language in its press releases or internal blogs are peppered with anachronisms and bias, it suggests their concern only goes so far.

“Companies are hungry to put actions behind their words,” said Writer founder

Read More Read more
women

How To Reduce Bias Where Research Says Black Women’s Natural Hair Impacts Interview Opportunities

Many studies explore how diversity in the workplace can create great opportunities for multiple viewpoints to be shared in ways that can strengthen company decision making from a holistic viewpoint. However, a recent research study conducted by Fuqua School of Business at Duke University suggests that there is evidence recommending that Black women with natural hairstyles are perceived as less professional than Black women with straightened hair. This evidence illuminates that discrimination against natural Black hairstyles is still prevalent in the workplace.

To determine if bias against natural haired Black women still exists in the workplace, the researchers recruited participants from various diasporas and asked them to role play as job recruiters screening candidates for employment. As participants role played, they were provided profiles of Black and White women job candidates and asked to rate various factors including their professionalism, competence and more. In the professionalism and competence categories,

Read More Read more
women

Women Founders of AI Startups Take Aim at Gender Bias

When Rana el Kaliouby was pitching investors in 2009 on her artificial-intelligence startup, Affectiva, she and her co-founder tried to steer clear of what she calls the “e-word”—emotion.

They were both women, and though their startup was designed to detect emotion in technology, they were sensitive to how they would be perceived. They feared they might not be taken seriously because “emotion” wasn’t in the traditional lexicon of many companies and funders—and because it carried female connotations in a largely male industry.

“We danced around it,” she says, adding that they called themselves a “sentiment” company instead. “Investors invest in what they know, and we were so different from what they were used to.”

A decade later, that is changing. Though the field of artificial intelligence remains heavily male-dominated, female leaders have made noticeable inroads. Some say they want to build workplaces that are more inviting to diverse workforces, and

Read More Read more