Google researcher Meredith Whittaker on the dangers of AI and brain-reading technology

Google Open Research lead
Meredith Whittaker
Photo by Doug Cody,
courtesy of O’Reilly Media

• Meredith Whittaker, a well-known AI researcher and
  ethicist, spoke at the Artificial Intelligence 2018 conference
  in San Francisco. 
• Whittaker painted a gloomy picture of a future where
  tech companies possess the ability to read and store human
  thoughts, and authorities have the ability to subpoena
  them.
• She said that now is the time to ensure the safe use of
  AI and other technologies.  
• One way she suggested that we do that is not to allow
  all this power to rest in the hands of a few privileged
  people.

Listening to Meredith Whittaker
speak about artificial intelligence and the future can chill the
blood, as she suggests dystopian scenarios where not even our
thoughts are safe.

As a research scientist at New
York University, and the leader of Google’s Open Research
Group
, Whittaker
studies the
social implications of AI.  At the
Artificial
Intelligence 2018 Conference in San Francisco on Thursday,
Whittaker warned thousands of attendees that the time to protect
ourselves from the misuse of AI and other new and formidable
technologies is now.

“A bit of a disclaimer,” said
Whittaker at the outset of her conference session. “This talk is
going to be dark.”

AI is one of the tech sector’s
hottest fields. All the big companies, including Google,
Facebook, IBM, Amazon, Microsoft, Apple and Intel, are spending
big on AI research and development. Those investments are
beginning to pay off as AI applications are already having a big
impact on industries including, but not limited to, travel,
healthcare and law enforcement. According to Whittaker, this is
just the start.  

To illustrate where technology is
going and the questions it will raise, Whittaker noted
how Facebook and Neuralink, a company co-founded and
led by Elon Musk, are both searching for the means to enable
humans to control computers and other devices solely with their
brains.  One approach would
attach tiny electrodes to the brain.

The technology could enable
people to control video games or mobile phones via their
thoughts.

But Whittaker said that our
thoughts could then conceivably be mapped, studied and eventually
interpreted. She asked the crowd to consider this: Musk, Facebook
CEO Mark Zuckerberg, and other tech titans could one day possess
the ability to read and store our thoughts.

Former
Google Cloud COO Diane Bryant
Greg
Sandoval/Business iInsider

“Your thoughts will belong to
them,” she said, adding that she considered such a prospect
“terrifying” and “creepy.”

Whittaker said that in world
where thoughts could be warehoused, she asked, what would happen
when one of the companies didn’t feel like fighting the
authorities when they “subpoenaed our thought logs?”

In a more immediate sense,
Whittaker highlighted how even now, in big and little ways, we
can find bias in how AI applications are written and applied. She
said that some facial recognition has been found to be biased
against people of color.

One example of this came in July,
when the American Civil Liberties Union said that Amazon’s facial
recognition software

misidentified 28
members

of Congress
during a trial of the software. The group also said that most of
the lawmakers incorrectly flagged were people of color.

Another incident, one that
Whittaker didn’t mention, occurred closer to home. A Google
teleconferencing system, designed to track a person’s face as
they spoke, had to be rolled back after it turned out the
system

 failed
to recognize

faces
belonging to people of color, said Diane Bryant, the former
Google Cloud exec, during a presentation last month.

At the root of the problem,
Whittaker suggests, is that all this powerful technology is
controlled by a relative few people, which she said were largely
white, male, affluent, and located in the Bay Area.

“The current boom is only
possible due to the increase in power of the tech industry,” she
said. “Globally, there is only seven or so companies capable of
creating AI at scale.”

She told the crowd: “We don’t
have the ability to realize the potential of AI without
recognizing it’s downsides.”