Mobile Menu - OpenMobile Menu - Closed

Connect

Twitter icon
Facebook icon
Flickr icon
Instagram icon

Congressman Ruben Gallego

Representing the 7th District of Arizona

32 Democrats Hold Santander Accountable for Potentially Discriminatory Employment Practices

November 15, 2017
Press Release

WASHINGTON, DC ­– Today, Reps. Ruben Gallego (AZ-07), Yvette Clarke (NY-09) and 30 of their House Democratic colleagues wrote to Scott Powell, Chief Executive Officer of Santander Consumer USA regarding the company’s use of automated call monitoring software to make employee evaluation and compensation decisions.

Automated software analytics – such as the Call Miner program used by Santander – are often built based on data sets that skew white, male, and toward certain dialects – which has been shown to result in telephone operators with accents, softer voices, and other varying speech patterns scoring poorly.

“We are concerned about the possibility that the use of speech recognition tools like Call Miner in worker evaluation and compensation could lead to bias against certain groups of workers on the basis of race, ethnicity, gender, or even regional dialect,” they wrote.

The letter asks Santander to take steps to ensure that the use of its call monitoring software is not having a discriminatory impact on the company’s workforce, including minority, female and disabled employees.

“If you have not, in fact, taken any specific precautionary measures or collected data to ensure that all groups of workers receive fair and equitable treatment, we strongly encourage you to suspend use of the Call Miner tool for purposes of evaluation and compensation until you have done so,” the letter said.

The full text of the letter is below and the signed letter can be found here.

 

Dear Mr. Powell:

As you likely know, we are deeply committed to combating discrimination of any form. In this context, we are writing out of concern about the potentially discriminatory implications of practices outlined in a recent report by the AFL-CIO and National Employment Law Project, "Wheeling and Dealing Misfortune: How Santander's high pressure tactics hurt workers and auto loan consumers."

This new report describes a system of intense monitoring of Santander Consumer workers' calls through an automated system known as “Call Miner.” According to the report, Call Miner “monitors and records employees’ calls and inspects their speech for potential problems.” The information gleaned from this monitoring is then used in workers’ performance evaluations, and may result in punitive actions including denial of bonuses to workers whose Call Miner scores are too low. We also understand that it is used as part of a “forced ranking” system that Santander uses to grade workers, which can ultimately lead to discipline or termination for low-ranked workers. Workers who recently came to brief Congress on conditions at Santander Consumer reported similar personal experiences.

We are concerned about the possibility that the use of speech recognition tools like Call Miner in worker evaluation and compensation could lead to bias against certain groups of workers on the basis of race, ethnicity, gender, or even regional dialect. Many studies and reports have shown that speech recognition software can fail to recognize different dialects, accents or tones. According to linguist researcher Rachael Tatman, Google’s speech recognition software is much more consistent with male voices than female voices.  Other reports note that “spoken accents… continue to perplex artificial intelligence.”  Tatman states that the groups that are underrepresented in data sets that inform speech recognition software tend to be groups that are marginalized more broadly in society.

If Call Miner’s software also fails to recognize certain groups’ speech patterns, it could result in discriminatory impacts on employment and compensation. Indeed, the AFL-CIO and NELP’s report cites one worker for whom Call Miner “was unable to identify the specific pronunciation of his words” and another with a lisp and a soft voice who observed that Call Miner was failing to understand her speech.

We ask that you inform us what, if any, precautions you have taken to ensure that the use of Call Miner does not have a discriminatory impact on your workforce. Please also provide us with any analysis that you have done to monitor whether or not protected classes of workers are treated fairly by the use of Call Miner. If you have not, in fact, taken any specific precautionary measures or collected data to ensure that all groups of workers receive fair and equitable treatment, we strongly encourage you to suspend use of the Call Miner tool for purposes of evaluation and compensation until you have done so.

Thanks in advance for your consideration and we look forward to your prompt reply.

Sincerely,

###