By Hugo Greenhalgh
LONDON, Aug 1 (Openly) - Artificial intelligence could help LGBT+ people avoid bias in job hiring but the technology is open to "profound abuse", self-described mad scientist Vivienne Ming said on Thursday.
The entrepreneur, who once built a computer program that could predict sexuality from a LinkedIn profile, said AI was the perfect tool to liberate - and trap - LGBT+ people.
"Finding difference is almost exactly what deep machine learning is perfect at," she told the Thomson Reuters Foundation in a telephone interview.
"But it's not the system (of collecting data) that should drive our concern. It's how it gets used."
Artificial intelligence is seen as the next frontier by computer scientists, raising hopes for an era when machines can perform tasks with the brain of a human.
But it has also raised fears that AI-fuelled machines might exacerbate the prejudices of their programmers.
Ming, an AI evangelist who has described herself as a "Professional Mad Scientist", sees high risk in letting impersonal algorithms loose on personal matters.
"(It's) wildly dangerous," Ming said, citing the need to balance individual civil liberties with the demands of large corporations or governments.
Ming sees clear risks of using the technology in a labour market where being different can cost.
Studies show that more than a third of LGBT+ people in Britain hide their identities at work.
Almost half do so in the United States.
Trained as a theoretical neuroscientist, Ming has put a price on bias, saying there was a "tax on being different" that coerces outsiders to spend years more on training so as to measure up to a mainstream candidate in the eyes of recruiters.
The "tax" on being gay in Britain can be between $65,000 and $77,000 over a lifetime, the 47-year-old estimated.
While AI is a "phenomenally powerful tool" to unpick these prejudices, "it's not magic", she added.
A BETTER PERSON
Ming shot to fame after a celebrated 2017 TEDx talk on "Making a Better Person", a vision of self-improvement.
Her own version of self-improvement has not been an easy ride since growing up as Evan Smith in Monterey, California.
Smith's early years were marked by academic excellence, before a growing sense of gender discomfort led him to drop out of university and sleep in his car.
The turnaround started with a return to academia, embarking on a psychology and theoretical neuroscience PhD in 2001 at Carnegie Mellon University in Pittsburgh, where he fell in love with Norma Chang.
In 2005, Smith told Chang that he felt that he was in fact a woman, later opting for gender-affirmation surgery.
They are now married with two children.
Yet Ming, which is an amalgam of their two surnames, said she looks back on those years with "a variety of lenses".
"One is crystal clear, and I don't use it to beat myself up - I use it to motivate myself," she said. "Given all the things I've invented, particularly in the healthcare space, how many people would be alive if I (had studied) 10 years earlier?"
In particular, Ming is referring to an AI-based system she devised with Chang to track their son's diabetes and help regulate his supply of insulin.
She went on to found Socos Labs, which calls itself "an independent think tank exploring the future of human potential".
Ming has also developed an early warning AI system to predict manic episodes in bipolar sufferers.
But ultimately, Ming wants her work on hiring and diversity to help others win the opportunities she missed.
"I fell out of the system and ended up on the streets," Ming said.
Recognising that even her background from a relatively comfortable family did not cushion that fall, "what does that mean for everyone else?"
"What world would we live in today if everyone had the same opportunity to do the type of work I do?" (Reporting by Hugo Greenhalgh @hugo_greenhalgh; Editing by Lyndsay Griffiths. Please credit the Thomson Reuters Foundation, the charitable arm of Thomson Reuters that covers humanitarian news, women's and LGBT+ rights, human trafficking, property rights, and climate change. Visit http://news.trust.org)
Openly is an initiative of the Thomson Reuters Foundation dedicated to impartial coverage of LGBT+ issues from around the world.
Our Standards: The Thomson Reuters Trust Principles.