Sexual Harassment

Advanced Discovery this week launched a sexual harassment taxonomy for its Riskcovery analytics suite that would allow companies to examine existing company data for potential sexual harassment concerns.

David Trachtenberg, chief marketing officer for Advanced Discovery, explained that companies often deal with these issues in litigation, despite having the capacity to screen for them prior to costly and exhaustive litigation. “Before data becomes evidence in litigation in traditional e-discovery practices, it's already sitting there with the company in their data before its highlighted as a potential evidence for litigation,” he noted.

Advanced Discovery's predictive risk assessment tools to date have focused on attempting to identify potential insider trading or compliance concerns. Collin Miller, director of Riskcovery services, noted that the company increasingly received calls from businesses looking to apply some of these same approaches to sexual harassment claims and Equal Employment Opportunity Commission (EEOC) violations.

Trachtenberg attributed some of this renewed attention to the number of high profile stories of workplace sexual misconduct following from the #MeToo campaign. “At the simplest level, it's a pervasive issue. It's not just coming out of Hollywood, it's coming out of board rooms as well,” Trachtenberg said.

Miller explained that the tool uses algorithms to scan through pre-process data sitting in email servers or mobile devices and look for potential red flags for inappropriate workplace conduct. However, because those red flags are a little more difficult to find than just keyword searching for “sexual harassment,” the Riskcovery team has done a little extra work to segment and identify recurring comments or behaviors that are likely to lead to or corroborate sexual harassment claims.

“That's not necessarily something you can just run keywords for. We're actually running a behavioral analysis,” Miller said.

Using machine learning to predict human behavior has been one of the more controversial uses of artificial intelligence over the last few years. Both scholars and high profile reporting by ProPublica and other media outlets have pointed to the potential for predictive algorithms to reinforce and recode existing bias and stereotypes in data.

Miller said Advanced Discovery takes a fairly “reactionary ingestive” approach to sexual harassment analysis to ensure that analysis is unbiased. The platform takes in any claim information provided about potential harassers, including descriptions of any misconduct or inappropriate behavior, and will use those as samples to examine potential conduct.

More importantly, Miller noted that the tool isn't a set-it-and-forget-it type platform, but rather an investigatory platform. “It goes different ways. Sometimes we end up proving the negative; other times we go in and find exactly what's going on,” he said.

The tool is also easily customized to particular business compliance standards, meaning that companies can calibrate the tool to identify specific behaviors that violate its internal policies.

Given the new spotlight and seriousness with which businesses are now handling sexual harassment allegations, Trachtenberg expects that tools like Riskcovery are likely to see a swift increase in usage.

“I think and I expect that we're going to start seeing a shift from the investigative 'let's go see if there's an issue and prove the positive or negative,' approach to broadly being more proactive on high-risk individuals like higher management,” he said.

“'We didn't know' doesn't pass muster anymore,” Trachtenberg added.