New #MeTooBots scan emails to detect harassment
New anti-harassment software has drawn eager customers, but critics are asking hard questions.
By Nathan Godin | Feb 11, 2020 | Print-friendly

Artificial-intelligence programmers are developing software programs that will scan read emails for content indicates sexual harassment or bullying. The programmers plan to introduce the software to businesses across the globe who are in the market for new solutions to help root out sexual harassment within their workplaces.

The software will flag any communications that its algorithms determine to be problematic. But even the developers admit that a computer may have a hard time knowing harassment when it sees it.

Jay Lieb, chief executive of NexLP, is marketing a platform that already has more than 50 corporate customers. Lieb said in an interview harassment comes in more forms than even he knew.

"I thought it was just talking dirty. It comes in so many different ways. It might be 15 messages it could be racy photos," he said.

He declined to say exactly how the software will determine which communications are harassing and which are not. In more general terms, it will look for anomalies in the language or in the frequency and timing of messages over spans of weeks, he said.

Critics question whether the software will be able to learn all of the cues that indicate harassment. Some critics also contend that electronically monitoring employees' communications creates a climate of mistrust and that companies will need to find ways to protect employees' privacy while collecting all of their data.

The law firm Morgan Lewis is using the technology in an alternative way: to analaze clients' past communications, according to Tess Blair, a partner at the firm. Blair said that while the technology has helped attorneys build cases, the attorneys are able to exercise their human judgement and determine which flagged messages are really relevant or not.